# Furkan Gözükara · SECourses > Furkan Gözükara (PhD computer engineer) builds and maintains 200+ 1-click generative AI apps, presets and video tutorials for Windows, Linux and cloud GPUs (RunPod, Massed Compute, SimplePod). Apps and installers are on Patreon (https://www.patreon.com/cw/SECourses); tutorials on YouTube (https://www.youtube.com/@SECourses). ## ComfyUI: 1-Click Installer, Presets and Tutorials - [LongCat Avatar Tutorial: Turn Your Photo and Voice Into a Talking Video](https://furkangozukara.com/apps/longcat-avatar-tutorial-turn-your-photo-and-voice-into-a-talking-video/): LongCat Avatar 1.6 turns one photo and one audio file into a lip-synced talking video on your own GPU; our 3-step preset runs 2.9x faster than 8 steps. - [ComfyUI Auto Installer: 1-Click Windows, Linux & Cloud With Presets](https://furkangozukara.com/apps/comfyui-auto-installer-1-click-windows-linux-cloud-with-presets/): ComfyUI auto installer that sets up Sage Attention, Triton, xFormers and Flash Attention in one click, with ready presets for image, video and lip sync models. - [MiniMax H3 Image: New Preset Fixes Tiny-Detail Errors at 16 MP](https://furkangozukara.com/apps/minimax-h3-image-new-preset-fixes-tiny-detail-errors-at-16-mp/): MiniMax H3 Image gets a new text-to-image preset that draws small objects such as watch gears, gem settings and tools more correctly at 16 MP in 6 steps. - [MiniMax H3 Image in ComfyUI: 16 MP Images and Edits in 6 Steps](https://furkangozukara.com/apps/minimax-h3-image-in-comfyui-16-mp-images-and-edits-in-6-steps/): MiniMax H3 Image is a new 33.1B model that generates and edits 16 MP pictures in 6 steps on your own GPU; our ComfyUI V150 ships two ready presets for it. - [Pika H3 Flex Presets for ComfyUI & SwarmUI: Video With Synced Sound](https://furkangozukara.com/apps/pika-h3-flex-presets-for-comfyui-swarmui-video-with-synced-sound/): Pika H3 Flex from Pika Labs now runs in ComfyUI and SwarmUI presets: a 5-second 1344 x 768 clip with stereo sound takes about a minute on an RTX 5090. - [ComfyUI & SwarmUI Update: Anima Character Transfer, MOSS-Image Control](https://furkangozukara.com/apps/comfyui-swarmui-update-anima-character-transfer-moss-image-control/): ComfyUI V146 and SwarmUI Model Downloader V186 add Anima one-image character transfer and MOSS-Image color and depth control, plus 91 SwarmUI presets in V85. - [Higgsfield Genjutsu Character Replace Locally: ComfyUI & SwarmUI](https://furkangozukara.com/apps/higgsfield-genjutsu-character-replace-locally-comfyui-swarmui/): MiniMax H3 Genjutsu presets replace the person in any video with your own character from one image, keeping the clip's motion, camera and original sound. - [ComfyUI Latest Updates: New Models, Presets and Improvements](https://furkangozukara.com/apps/comfyui-latest-updates-new-models-presets-and-improvements/): ComfyUI update log with the latest four releases; V144 adds ten presets for Fibo 2.0, MOSS Image Edit and FastH3 Omni, for 106 ready presets in total. - [ComfyUI Install & FLUX 3 Klein: The Complete Beginner Tutorial](https://furkangozukara.com/apps/comfyui-install-flux-3-klein-the-complete-beginner-tutorial/): ComfyUI beginner tutorial: install on Windows and learn FLUX 3 Klein with 16 core examples, precise image edits, RTX 5090 tests and launcher tuning. - [ComfyUI Improvements for MiniMax H3 Video Continuation and LTX 2.5](https://furkangozukara.com/apps/comfyui-improvements-for-minimax-h3-video-continuation-and-ltx-2-5/): ComfyUI V131 lets MiniMax H3 continue videos from the final 1 to 56 frames with its native clip guide, and V129 adds 8 official LTX 2.5 presets with audio. - [MiniMax H3 Infinite Video: Make 2-Hour Movies in ComfyUI (0-Shot)](https://furkangozukara.com/apps/minimax-h3-infinite-video-make-2-hour-movies-in-comfyui-0-shot/): MiniMax H3 turns a folder of scene prompts into one long, coherent AI movie in ComfyUI, locally and 0-shot, with every clip merged automatically. - [Ideogram 4 JSON Prompting Tutorial With Ultimate Image Captioner Pro](https://furkangozukara.com/apps/ideogram-4-json-prompting-tutorial-with-ultimate-image-captioner-pro/): Ideogram 4 runs locally in SwarmUI and ComfyUI with Turbo, Balanced and Highest Quality presets; the masterclass teaches accurate JSON prompts. - [LTX 2 Ultimate Tutorial: ComfyUI & SwarmUI on Windows and Cloud](https://furkangozukara.com/apps/ltx-2-ultimate-tutorial-comfyui-swarmui-on-windows-and-cloud/): LTX2 video tutorial covering ComfyUI and SwarmUI setup, model downloads and ready presets plus Z-Image Base, on GPUs with as low as 6 GB VRAM. - [BF16 vs GGUF Q8 vs FP8 Scaled vs NVFP4: Quality and Speed Comparison](https://furkangozukara.com/apps/bf16-vs-gguf-q8-vs-fp8-scaled-vs-nvfp4-quality-and-speed-comparison/): BF16, GGUF Q8, FP8 Scaled and NVFP4 precisions compared for speed and quality in ComfyUI with CUDA 13; NVFP4 ran 87% faster than GGUF Q8 on Z Image Turbo. - [NVFP4 With CUDA 13 Tutorial: 100%+ Speed Gain & Quality Comparison](https://furkangozukara.com/apps/nvfp4-with-cuda-13-tutorial-100-speed-gain-quality-comparison/): NVFP4 models run over 100% faster in ComfyUI and SwarmUI with CUDA 13; this full tutorial compares NVFP4 quality on FLUX 2, Z Image Turbo and FLUX 1 in grids. - [SwarmUI to ComfyUI in 1-Click Tutorial: 40+ Presets, Shared Models](https://furkangozukara.com/apps/swarmui-to-comfyui-in-1-click-tutorial-40-presets-shared-models/): SwarmUI exports any preset as a ComfyUI workflow in 1 click; this tutorial runs Z Image Turbo with ControlNet and Wan 2.2 image-to-video presets in ComfyUI. - [ComfyUI & SwarmUI on Massed Compute and RunPod: Cloud GPU Tutorial](https://furkangozukara.com/apps/comfyui-swarmui-on-massed-compute-and-runpod-cloud-gpu-tutorial/): ComfyUI and SwarmUI set up in 1 click on Massed Compute and RunPod cloud GPUs, with multi-GPU backends for faster Qwen, FLUX and Wan 2.2 generation. - [MultiTalk Levelled Up: New ComfyUI Workflows for Better Animation](https://furkangozukara.com/apps/multitalk-levelled-up-new-comfyui-workflows-for-better-animation/): MultiTalk turns a static image and audio into talking and singing videos; 4 new ComfyUI workflow categories give better animation at the same VRAM and speed. - [ComfyUI & SwarmUI on RunPod Tutorial: 1-Click Setup With RTX 5000 GPUs](https://furkangozukara.com/apps/comfyui-swarmui-on-runpod-tutorial-1-click-setup-with-rtx-5000-gpus/): ComfyUI and SwarmUI on RunPod, step by step: install both with 1 click, connect ComfyUI as the SwarmUI backend and generate FLUX images and Wan 2.1 videos. - [ComfyUI Clothing Transfer & 3D Figure Workflows: 1-Click Tutorial](https://furkangozukara.com/apps/comfyui-clothing-transfer-3d-figure-workflows-1-click-tutorial/): ComfyUI workflows that transfer any clothing onto a new person and turn any person into a 3D toy figure, installed in 1 click and run with FLUX Dev models. - [SwarmUI With ComfyUI Backend Tutorial: Max Speed Install on Windows](https://furkangozukara.com/apps/swarmui-with-comfyui-backend-tutorial-max-speed-install-on-windows/): SwarmUI and ComfyUI Windows tutorial: properly install ComfyUI in a VENV with 1 click, then use it as the SwarmUI backend for maximum performance. ## SwarmUI: 1-Click Installer, Model Downloader and Tutorials - [LongCat Avatar Tutorial: Turn Your Photo and Voice Into a Talking Video](https://furkangozukara.com/apps/longcat-avatar-tutorial-turn-your-photo-and-voice-into-a-talking-video/): LongCat Avatar 1.6 turns one photo and one audio file into a lip-synced talking video on your own GPU; our 3-step preset runs 2.9x faster than 8 steps. - [SwarmUI Auto Installer & Ultimate Image and Video Model Downloader](https://furkangozukara.com/apps/swarmui-auto-installer-ultimate-image-and-video-model-downloader/): SwarmUI auto installer with extensions plus a unified AI models downloader for SwarmUI, ComfyUI, Automatic1111 and Forge, with 105 ready SwarmUI presets. - [MiniMax H3 Image: New Preset Fixes Tiny-Detail Errors at 16 MP](https://furkangozukara.com/apps/minimax-h3-image-new-preset-fixes-tiny-detail-errors-at-16-mp/): MiniMax H3 Image gets a new text-to-image preset that draws small objects such as watch gears, gem settings and tools more correctly at 16 MP in 6 steps. - [SwarmUI Full Tutorial: 1-Click Install, 4 MP Images, HD Video & Lip Sync](https://furkangozukara.com/apps/swarmui-full-tutorial-1-click-install-4-mp-images-hd-video-lip-sync/): SwarmUI full tutorial from a fresh Windows install: our installers, model downloader with 517+ models and 86 presets turn your PC into a complete AI studio. - [Pika H3 Flex Presets for ComfyUI & SwarmUI: Video With Synced Sound](https://furkangozukara.com/apps/pika-h3-flex-presets-for-comfyui-swarmui-video-with-synced-sound/): Pika H3 Flex from Pika Labs now runs in ComfyUI and SwarmUI presets: a 5-second 1344 x 768 clip with stereo sound takes about a minute on an RTX 5090. - [ComfyUI & SwarmUI Update: Anima Character Transfer, MOSS-Image Control](https://furkangozukara.com/apps/comfyui-swarmui-update-anima-character-transfer-moss-image-control/): ComfyUI V146 and SwarmUI Model Downloader V186 add Anima one-image character transfer and MOSS-Image color and depth control, plus 91 SwarmUI presets in V85. - [Higgsfield Genjutsu Character Replace Locally: ComfyUI & SwarmUI](https://furkangozukara.com/apps/higgsfield-genjutsu-character-replace-locally-comfyui-swarmui/): MiniMax H3 Genjutsu presets replace the person in any video with your own character from one image, keeping the clip's motion, camera and original sound. - [SwarmUI Latest Updates: Fibo 2.0, FastH3 Omni and New Presets](https://furkangozukara.com/apps/swarmui-latest-updates-fibo-2-0-fasth3-omni-and-new-presets/): SwarmUI update log with the four latest releases; V184 adds the Fibo 2.0, MOSS Image Edit and FastH3 Omni models and grows Amazing SwarmUI Presets to 84. - [SwarmUI Update Log: LTX 2.5 Presets, MiniMax H3 Continuation & More](https://furkangozukara.com/apps/swarmui-update-log-ltx-2-5-presets-minimax-h3-continuation-more/): SwarmUI latest updates: V172 adds multi-frame MiniMax H3 video continuation, and V170 brings 6 LTX 2.5 video presets with two-stage 1080p output and audio. - [Ideogram 4 JSON Prompting Tutorial With Ultimate Image Captioner Pro](https://furkangozukara.com/apps/ideogram-4-json-prompting-tutorial-with-ultimate-image-captioner-pro/): Ideogram 4 runs locally in SwarmUI and ComfyUI with Turbo, Balanced and Highest Quality presets; the masterclass teaches accurate JSON prompts. - [LTX 2 Ultimate Tutorial: ComfyUI & SwarmUI on Windows and Cloud](https://furkangozukara.com/apps/ltx-2-ultimate-tutorial-comfyui-swarmui-on-windows-and-cloud/): LTX2 video tutorial covering ComfyUI and SwarmUI setup, model downloads and ready presets plus Z-Image Base, on GPUs with as low as 6 GB VRAM. - [NVFP4 With CUDA 13 Tutorial: 100%+ Speed Gain & Quality Comparison](https://furkangozukara.com/apps/nvfp4-with-cuda-13-tutorial-100-speed-gain-quality-comparison/): NVFP4 models run over 100% faster in ComfyUI and SwarmUI with CUDA 13; this full tutorial compares NVFP4 quality on FLUX 2, Z Image Turbo and FLUX 1 in grids. - [SwarmUI to ComfyUI in 1-Click Tutorial: 40+ Presets, Shared Models](https://furkangozukara.com/apps/swarmui-to-comfyui-in-1-click-tutorial-40-presets-shared-models/): SwarmUI exports any preset as a ComfyUI workflow in 1 click; this tutorial runs Z Image Turbo with ControlNet and Wan 2.2 image-to-video presets in ComfyUI. - [Qwen Image Edit 2511 Tutorial: 12-Case Comparison With 2509](https://furkangozukara.com/apps/qwen-image-edit-2511-tutorial-12-case-comparison-with-2509/): Qwen Image Edit 2511 is a free, open source image editor that challenges Nano Banana Pro, with native 2560x2560 output in 12 steps, even on 6 GB GPUs. - [Z-Image Turbo Tutorial: LoRA Training, ControlNet Union & Presets](https://furkangozukara.com/apps/z-image-turbo-tutorial-lora-training-controlnet-union-presets/): Z-Image Turbo step by step: generate with ready SwarmUI presets, add Fun ControlNet Union and train LoRAs with Ostris AI Toolkit on GPUs from 6 GB. - [Qwen Image Next Level Realism: LoRA Workflow Tutorial, 4+4 Steps](https://furkangozukara.com/apps/qwen-image-next-level-realism-lora-workflow-tutorial-4-4-steps/): Qwen Image presets for SwarmUI with new realism LoRAs reach next level realism in only 4+4 steps; the full tutorial also covers inpainting and outpainting. - [ComfyUI & SwarmUI on Massed Compute and RunPod: Cloud GPU Tutorial](https://furkangozukara.com/apps/comfyui-swarmui-on-massed-compute-and-runpod-cloud-gpu-tutorial/): ComfyUI and SwarmUI set up in 1 click on Massed Compute and RunPod cloud GPUs, with multi-GPU backends for faster Qwen, FLUX and Wan 2.2 generation. - [SwarmUI and ComfyUI Tutorial: Free Local AI Image Generation](https://furkangozukara.com/apps/swarmui-and-comfyui-tutorial-free-local-ai-image-generation/): SwarmUI with a ComfyUI backend gives free, local AI image generation: a 1-click installer, realism and style presets, text-based editing and 4K/8K upscaling. - [Qwen Image Edit Full Tutorial: 26 Demo Cases, Prompts & Images](https://furkangozukara.com/apps/qwen-image-edit-full-tutorial-26-demo-cases-prompts-images/): Qwen Image Edit full tutorial with 26 different demo cases, from photo restoration and text changes to stylization, object removal and outpainting in SwarmUI. - [Wan 2.2, FLUX Krea & Qwen Image Upgraded: Fast 8-Step Presets Tutorial](https://furkangozukara.com/apps/wan-2-2-flux-krea-qwen-image-upgraded-fast-8-step-presets-tutorial/): Wan 2.2, Qwen Image, FLUX and FLUX Krea get new SwarmUI presets after a week of research, and new 8-step LoRAs make Wan 2.2 video and Qwen Image fast. - [Wan 2.2 & FLUX Krea Full Tutorial: 1-Click Install & SwarmUI Presets](https://furkangozukara.com/apps/wan-2-2-flux-krea-full-tutorial-1-click-install-swarmui-presets/): Wan 2.2 and FLUX Krea Dev in SwarmUI with a ComfyUI backend, using ready presets built from hundreds of parameter tests for videos from text or images. - [FLUX Kontext Dev Tutorial: Local Image Editing on Windows](https://furkangozukara.com/apps/flux-kontext-dev-tutorial-local-image-editing-on-windows/): FLUX Kontext Dev edits any part of an image with just a prompt, no mask or ControlNet, and can restore, colorize, outpaint and combine images in SwarmUI. - [Wan 2.1 FusionX & Self Forcing LoRA: 8-Step Video + FLUX Upscaling Guide](https://furkangozukara.com/apps/wan-2-1-fusionx-self-forcing-lora-8-step-video-flux-upscaling-guide/): Wan 2.1 FusionX and Self Forcing LoRA generate local videos from text or images in only 8 steps with our SwarmUI presets, plus FLUX 2x latent upscaling. - [SwarmUI Beginner's Guide: Generate AI Videos With Wan 2.1 and More](https://furkangozukara.com/apps/swarmui-beginner-s-guide-generate-ai-videos-with-wan-2-1-and-more/): SwarmUI beginner's guide to AI video by its developer mcmonkey: pick a model such as Wan 2.1 14B, add it to the right folder and generate your first clips. - [ComfyUI & SwarmUI on RunPod Tutorial: 1-Click Setup With RTX 5000 GPUs](https://furkangozukara.com/apps/comfyui-swarmui-on-runpod-tutorial-1-click-setup-with-rtx-5000-gpus/): ComfyUI and SwarmUI on RunPod, step by step: install both with 1 click, connect ComfyUI as the SwarmUI backend and generate FLUX images and Wan 2.1 videos. - [SwarmUI Teacache Tutorial: Wan 2.1 I2V & T2V Presets, 2x Faster](https://furkangozukara.com/apps/swarmui-teacache-tutorial-wan-2-1-i2v-t2v-presets-2x-faster/): SwarmUI Teacache full tutorial: speed up Wan 2.1 image-to-video and text-to-video 2x and more with ready presets, using ComfyUI as the backend. - [Wan 2.1 T2V & I2V in SwarmUI Tutorial: CausVid LoRA for Extreme Speed](https://furkangozukara.com/apps/wan-2-1-t2v-i2v-in-swarmui-tutorial-causvid-lora-for-extreme-speed/): Wan 2.1 text-to-video and image-to-video in SwarmUI with the CausVid LoRA at just 8 steps; on an RTX 5090 a clip took 90 seconds versus 170 on an RTX 3090 Ti. - [SwarmUI With ComfyUI Backend Tutorial: Max Speed Install on Windows](https://furkangozukara.com/apps/swarmui-with-comfyui-backend-tutorial-max-speed-install-on-windows/): SwarmUI and ComfyUI Windows tutorial: properly install ComfyUI in a VENV with 1 click, then use it as the SwarmUI backend for maximum performance. - [SwarmUI Free Kaggle Notebook: FLUX, SDXL and SD 3.5 on Dual T4 GPUs](https://furkangozukara.com/apps/swarmui-free-kaggle-notebook-flux-sdxl-and-sd-3-5-on-dual-t4-gpus/): SwarmUI notebook for a free Kaggle account that runs FLUX Dev Q8 GGUF on both T4 GPUs at once, generating 2 images at the same time, with a video tutorial. - [FLUX Tools SwarmUI Tutorial: Inpainting, Outpainting, Redux & Canny](https://furkangozukara.com/apps/flux-tools-swarmui-tutorial-inpainting-outpainting-redux-canny/): FLUX.1 Tools complete SwarmUI tutorial: inpaint and outpaint with FLUX Fill, restyle and combine images with Redux, and guide images with Depth and Canny. - [SD 3.5 Large vs FLUX Dev: Best Sampler, Scheduler & FP8 Comparison](https://furkangozukara.com/apps/sd-3-5-large-vs-flux-dev-best-sampler-scheduler-fp8-comparison/): Stable Diffusion 3.5 Large tested with 62 prompts to find the best sampler and scheduler, plus an SD 3.5 Large vs FLUX Dev comparison with raw grids shared. ## LoRA and Fine-Tuning Training: Kohya, Musubi Tuner, OneTrainer, AI Toolkit - [Musubi Tuner App: 1-Click LoRA & Fine-Tune Training With 435 Presets](https://furkangozukara.com/apps/musubi-tuner-app-1-click-lora-fine-tune-training-with-435-presets/): SECourses Musubi Tuner trains LoRAs and full fine-tunes of Qwen Image, Wan 2.2, FLUX 2, Z Image and LTX 2.3 on Windows or cloud GPUs with 435 ready presets. - [Ostris AI-Toolkit 1-Click Installers: LoRA Training on Windows & Cloud](https://furkangozukara.com/apps/ostris-ai-toolkit-1-click-installers-lora-training-on-windows-cloud/): Ostris AI-Toolkit installs automatically on Windows, RunPod, SimplePod and Massed Compute and is now patched to train on custom single-file base models. - [Qwen Image 2512 Training Comparison: The Best Model to Train](https://furkangozukara.com/apps/qwen-image-2512-training-comparison-the-best-model-to-train/): Qwen Image 2512 is the new king for training: the video compares trained Base, 2512, Edit 2509 and Edit 2511 models and shows a big leap in realism. - [Qwen Image 2512 First Training: 120 vs 180 vs 240 Epochs Comparison](https://furkangozukara.com/apps/qwen-image-2512-first-training-120-vs-180-vs-240-epochs-comparison/): Qwen Image 2512 first training results: the model behaves like FLUX Dev, and more training makes it more realistic with less noise on a 28-image dataset. - [Wan 2.2 LoRA Training Tutorial: Text to Image & Video, Windows & Cloud](https://furkangozukara.com/apps/wan-2-2-lora-training-tutorial-text-to-image-video-windows-cloud/): Wan 2.2 complete LoRA training tutorial with Musubi Tuner, based on over 64 trainings; the ready configs train locally on GPUs with as little as 6 GB VRAM. - [Z-Image Turbo Tutorial: LoRA Training, ControlNet Union & Presets](https://furkangozukara.com/apps/z-image-turbo-tutorial-lora-training-controlnet-union-presets/): Z-Image Turbo step by step: generate with ready SwarmUI presets, add Fun ControlNet Union and train LoRAs with Ostris AI Toolkit on GPUs from 6 GB. - [Kohya FLUX LoRA Training: Full Tutorial & Configs for Windows & Cloud](https://furkangozukara.com/apps/kohya-flux-lora-training-full-tutorial-configs-for-windows-cloud/): Kohya SS GUI FLUX LoRA training step by step, with the best optimized configs for 4 GB to 48 GB GPUs and a 68 minute Windows tutorial with 74 chapters. - [Kohya FLUX Fine Tuning / DreamBooth Training Tutorial: Windows & Cloud](https://furkangozukara.com/apps/kohya-flux-fine-tuning-dreambooth-training-tutorial-windows-cloud/): Kohya FLUX full fine tuning / DreamBooth trains full checkpoints with configs for GPUs with as low as 8 GB VRAM, all using Torch Compile on Windows and Linux. - [FLUX 2 vs FLUX SRPO & New Kohya SS GUI Premium FLUX Training App](https://furkangozukara.com/apps/flux-2-vs-flux-srpo-new-kohya-ss-gui-premium-flux-training-app/): FLUX 2 Pro compared with a trained FLUX SRPO model, plus the new Kohya trainer V35 that trains FLUX on GPUs with as little as 6 GB VRAM. - [Qwen Image vs FLUX SRPO Training Comparison: 20 Images, Same Dataset](https://furkangozukara.com/apps/qwen-image-vs-flux-srpo-training-comparison-20-images-same-dataset/): Qwen Image base model vs FLUX SRPO training on the same 28-image dataset: 20 side-by-side results show Qwen far ahead at complex prompts and emotions. - [Qwen Image Training Tutorial: LoRA & Fine Tuning, Base & Edit Models](https://furkangozukara.com/apps/qwen-image-training-tutorial-lora-fine-tuning-base-edit-models/): Qwen Image LoRA and full fine-tuning tutorial with SECourses Musubi Tuner, built on 21 days of research; trains on Windows GPUs with as little as 6 GB VRAM. - [Claude AI Edits Word & Excel Files, Plus Wan Training Arrives](https://furkangozukara.com/apps/claude-ai-edits-word-excel-files-plus-wan-training-arrives/): Claude AI now edits Word documents and Excel spreadsheets; the video also shows Wan and Qwen Image training in Musubi Tuner and IndexTTS2 voice cloning. - [Qwen Image LoRA Training: Stage 1 Results and Pre-Made Configs](https://furkangozukara.com/apps/qwen-image-lora-training-stage-1-results-and-pre-made-configs/): Qwen Image LoRA results from more than 35 Stage 1 trainings with SECourses Musubi Tuner; the published pre-made configs train on GPUs with as low as 6 GB. - [Decoding the Timestep Shift in Diffusion Training: Qwen Image & FLUX](https://furkangozukara.com/apps/decoding-the-timestep-shift-in-diffusion-training-qwen-image-flux/): Timestep sampling shift explained for diffusion models: Qwen Image uses an exponential shift and FLUX a linear one, and 2.205 matches Qwen at 1664x928. - [Kohya FLUX DreamBooth Results: 20 Profile Images Generated Locally](https://furkangozukara.com/apps/kohya-flux-dreambooth-results-20-profile-images-generated-locally/): Kohya FLUX DreamBooth showcase: 20 profile images generated recently in SwarmUI to change my profile photo, each 2x latent upscaled to 4 megapixels. - [Kohya vs OneTrainer for FLUX Dev Fine-Tuning / DreamBooth: Comparison](https://furkangozukara.com/apps/kohya-vs-onetrainer-for-flux-dev-fine-tuning-dreambooth-comparison/): Kohya and OneTrainer compared for FLUX Dev fine-tuning: OneTrainer with EMA and a copied Kohya config gave weaker results than the best Kohya config. - [Kohya GUI: Convert FLUX, SD 3.5 & SDXL LoRAs to FP8 to Save Space](https://furkangozukara.com/apps/kohya-gui-convert-flux-sd-3-5-sdxl-loras-to-fp8-to-save-space/): Converting a trained LoRA to FP8 with Kohya GUI shrinks a 2.5 GB FP32 file to 634 MB with almost the best quality, as the FP8 vs FP16 vs FP32 grid shows. - [FLUX Fine-Tuning / DreamBooth Images: 2x Latent Upscale, 2048x2048](https://furkangozukara.com/apps/flux-fine-tuning-dreambooth-images-2x-latent-upscale-2048x2048/): FLUX fine-tuning / DreamBooth images made with a simple prompt and 2x latent upscale; each is 2048x2048 and training works on as low as 6 GB GPUs. - [Kohya FLUX LoRA on 4 GB and DreamBooth / Fine-Tuning on 6 GB GPUs](https://furkangozukara.com/apps/kohya-flux-lora-on-4-gb-and-dreambooth-fine-tuning-on-6-gb-gpus/): Kohya now trains FLUX LoRAs on 4 GB GPUs and full FLUX DreamBooth / fine-tuning on 6 GB, thanks to faster block swapping, FP8 and training only certain layers. - [FLUX Fine-Tuning vs LoRA: Example Dataset, Trained Models & Configs](https://furkangozukara.com/apps/flux-fine-tuning-vs-lora-example-dataset-trained-models-configs/): FLUX fine-tuning / DreamBooth vs LoRA experiment shared in full: a 28-image dataset, Kohya configs, grids and every trained checkpoint free on Hugging Face. - [Perfect Quality Training Image Dataset for FLUX, SD 3.5, SDXL & SD 1.5](https://furkangozukara.com/apps/perfect-quality-training-image-dataset-for-flux-sd-3-5-sdxl-sd-1-5/): Example model training dataset of 28 maximum-quality 1024x1024 images, prepared to learn and test training, sized for the FLUX batch size 7 config. - [FLUX De-Distilled Fine-Tuning & LoRA Tests: Bleeding and CFG Results](https://furkangozukara.com/apps/flux-de-distilled-fine-tuning-lora-tests-bleeding-and-cfg-results/): FLUX de-distilled models train best at guidance scale 3.5 instead of 1, yet these DreamBooth and LoRA experiments show they still do not fix bleeding. - [FLUX DreamBooth Checkpoints: Convert 24 GB FP16 to 12 GB FP8 Tutorial](https://furkangozukara.com/apps/flux-dreambooth-checkpoints-convert-24-gb-fp16-to-12-gb-fp8-tutorial/): FLUX fine tuning / DreamBooth checkpoints take 23.8 GB in FP16; Kohya GUI converts them to 11.9 GB FP8 with no visible quality loss in the grids. - [FLUX LoRA vs Fine Tuning / DreamBooth Comparison: Batch Size 1 vs 7](https://furkangozukara.com/apps/flux-lora-vs-fine-tuning-dreambooth-comparison-batch-size-1-vs-7/): FLUX LoRA and Fine Tuning / DreamBooth compared with 15 and 256 training images at batch size 1 and 7, for both realism and stylization. - [Kohya GUI: Full FLUX Dev DreamBooth on 6 GB GPUs Without Quality Loss](https://furkangozukara.com/apps/kohya-gui-full-flux-dev-dreambooth-on-6-gb-gpus-without-quality-loss/): Kohya GUI can now fully fine-tune FLUX Dev on 6 GB GPUs with the same quality as 48 GB GPUs, and fine-tuning gives better results than any LoRA training. - [Multi-GPU FLUX Full Fine-Tuning Test on RunPod: 2x A100 80 GB Results](https://furkangozukara.com/apps/multi-gpu-flux-full-fine-tuning-test-on-runpod-2x-a100-80-gb-results/): Multi-GPU FLUX full fine-tuning / DreamBooth on 2x A100 80 GB GPUs gave only a 26% speed gain over one GPU, so 2x GPU training is not worth it yet. - [FLUX LoRA Single Block Training Research: Network Alpha & Rank Tests](https://furkangozukara.com/apps/flux-lora-single-block-training-research-network-alpha-rank-tests/): FLUX LoRA single block / layer training tested in Kohya GUI: one block needed over 10 times the usual learning rate and still gave sub-par results. - [FLUX Full Fine Tuning vs LoRA Comparison: Less Overfitting & Bleeding](https://furkangozukara.com/apps/flux-full-fine-tuning-vs-lora-comparison-less-overfitting-bleeding/): FLUX full fine tuning / DreamBooth gives way better results than LoRA training, as expected, with much less overfitting and bleeding. - [FLUX LoRA Trained on 256 Images of Myself: Mind-Blowing Results](https://furkangozukara.com/apps/flux-lora-trained-on-256-images-of-myself-mind-blowing-results/): FLUX LoRA trained with Kohya GUI on 256 phone photos of myself, including the Clip-L and T5-XXL text encoders; it learned body shape and fine facial details. - [FLUX LoRA Training With 256 Images: Full Workflow & Conclusions](https://furkangozukara.com/apps/flux-lora-training-with-256-images-full-workflow-conclusions/): FLUX LoRA training experiments with 256 images on 8x A6000 and 1x L40S GPUs: 80 epochs worked best, and doubling the learning rate for 8 GPUs overtrained. - [Kohya FLUX LoRA Training Experiments: All 104 Trainings Listed](https://furkangozukara.com/apps/kohya-flux-lora-training-experiments-all-104-trainings-listed/): FLUX LoRA training experiments run with Kohya on an 8x RTX A6000 RunPod machine, 3000 steps each, with every trained model listed and shared on Hugging Face. - [FLUX LoRA Training Tutorial: Kohya GUI on Windows and Cloud GPUs](https://furkangozukara.com/apps/flux-lora-training-tutorial-kohya-gui-on-windows-and-cloud-gpus/): FLUX LoRA training with Kohya GUI, distilled from 104 trainings into ranked configs for 8 GB to 48 GB GPUs, plus Windows and cloud tutorials. - [SECourses 3D Render Style FLUX LoRA Published on CivitAI](https://furkangozukara.com/apps/secourses-3d-render-style-flux-lora-published-on-civitai/): SECourses 3D Render FLUX LoRA model achieves style consistency; download it on CivitAI and get the full workflow with experiment results on Hugging Face. - [Training a Public FLUX Style LoRA: Captions vs No Captions Test](https://furkangozukara.com/apps/training-a-public-flux-style-lora-captions-vs-no-captions-test/): FLUX style LoRA training in public: captions vs no captions on a 114-image 3D render dataset, 500 epochs on 4x A6000 GPUs with Kohya. - [SDXL DreamBooth / Fine Tuning: Best Kohya Configs From 6 GB GPUs](https://furkangozukara.com/apps/sdxl-dreambooth-fine-tuning-best-kohya-configs-from-6-gb-gpus/): SDXL DreamBooth / fine tuning configs for Kohya SS GUI, based on results of 100+ full trainings, train on GPUs with as low as 6 GB VRAM via block swapping. - [SDXL Fine Tuning: 20 New OneTrainer Tests, Results and Better Workflow](https://furkangozukara.com/apps/sdxl-fine-tuning-20-new-onetrainer-tests-results-and-better-workflow/): SDXL fine tuning with OneTrainer on RealVis XL 4: 20 tests compare batch sizes, learning rates, xFormers, regularization images and loss weight functions. - [SDXL Training: AdamW Batch Size 30 vs Adafactor Batch Size 1 Comparison](https://furkangozukara.com/apps/sdxl-training-adamw-batch-size-30-vs-adafactor-batch-size-1-comparison/): SDXL training test on an RTX A6000: AdamW at batch size 30 vs the SECourses Adafactor config at batch size 1, which gave far better resemblance. - [LoRA Training Tips: Learning Rate, Network Rank, Alpha & Batch Size](https://furkangozukara.com/apps/lora-training-tips-learning-rate-network-rank-alpha-batch-size/): Excellent tips for LoRA training, slightly improved: how network rank, alpha, optimizer, batch size and precision change the effective learning rate. - [Extract LoRA From FLUX & SDXL DreamBooth Models: Tutorial & Comparison](https://furkangozukara.com/apps/extract-lora-from-flux-sdxl-dreambooth-models-tutorial-comparison/): LoRA extraction from FLUX and SDXL DreamBooth models in Kohya, compared with fine tuning and LoRA training; higher ranks, up to 640, give better results. - [OneTrainer SDXL Fine-Tuning Best Presets: DreamBooth on 8.5 GB VRAM](https://furkangozukara.com/apps/onetrainer-sdxl-fine-tuning-best-presets-dreambooth-on-8-5-gb-vram/): OneTrainer Stable Diffusion XL (SDXL) fine-tuning presets and installers; three presets with the same quality, from 8.5 GB VRAM to the fastest 17 GB. - [SDXL DreamBooth Captioning Comparison: WD14 vs Kosmos-2 vs Ohwx Man](https://furkangozukara.com/apps/sdxl-dreambooth-captioning-comparison-wd14-vs-kosmos-2-vs-ohwx-man/): Image captioning in SDXL fine-tuning / DreamBooth of a single person reduces likeness with almost no benefit, in this OneTrainer test using 10.3 GB VRAM. - [Stable Diffusion Regularization Images: 4K Man & Woman Class Dataset](https://furkangozukara.com/apps/stable-diffusion-regularization-images-4k-man-woman-class-dataset/): Stable Diffusion regularization images: 5200 hand-checked real man and 5200 woman photos, cropped and sorted by face quality for SDXL and SD training. ## AI Video Generation and Animation Apps - [LongCat Avatar Tutorial: Turn Your Photo and Voice Into a Talking Video](https://furkangozukara.com/apps/longcat-avatar-tutorial-turn-your-photo-and-voice-into-a-talking-video/): LongCat Avatar 1.6 turns one photo and one audio file into a lip-synced talking video on your own GPU; our 3-step preset runs 2.9x faster than 8 steps. - [SwarmUI Full Tutorial: 1-Click Install, 4 MP Images, HD Video & Lip Sync](https://furkangozukara.com/apps/swarmui-full-tutorial-1-click-install-4-mp-images-hd-video-lip-sync/): SwarmUI full tutorial from a fresh Windows install: our installers, model downloader with 517+ models and 86 presets turn your PC into a complete AI studio. - [Henge 1-Click Video Studio: Character Swap, Motion Transfer, Head Swap](https://furkangozukara.com/apps/henge-1-click-video-studio-character-swap-motion-transfer-head-swap/): Henge uses MiniMax H3 to swap characters, transfer motion, swap heads and sharpen videos; a 12 second Character Swap took about 4 minutes on an RTX 5090. - [Pika H3 Flex Presets for ComfyUI & SwarmUI: Video With Synced Sound](https://furkangozukara.com/apps/pika-h3-flex-presets-for-comfyui-swarmui-video-with-synced-sound/): Pika H3 Flex from Pika Labs now runs in ComfyUI and SwarmUI presets: a 5-second 1344 x 768 clip with stereo sound takes about a minute on an RTX 5090. - [Higgsfield Genjutsu Character Replace Locally: ComfyUI & SwarmUI](https://furkangozukara.com/apps/higgsfield-genjutsu-character-replace-locally-comfyui-swarmui/): MiniMax H3 Genjutsu presets replace the person in any video with your own character from one image, keeping the clip's motion, camera and original sound. - [ComfyUI Improvements for MiniMax H3 Video Continuation and LTX 2.5](https://furkangozukara.com/apps/comfyui-improvements-for-minimax-h3-video-continuation-and-ltx-2-5/): ComfyUI V131 lets MiniMax H3 continue videos from the final 1 to 56 frames with its native clip guide, and V129 adds 8 official LTX 2.5 presets with audio. - [MiniMax H3 Infinite Video: Make 2-Hour Movies in ComfyUI (0-Shot)](https://furkangozukara.com/apps/minimax-h3-infinite-video-make-2-hour-movies-in-comfyui-0-shot/): MiniMax H3 turns a folder of scene prompts into one long, coherent AI movie in ComfyUI, locally and 0-shot, with every clip merged automatically. - [Wan2GP 1-Click Installer for Windows, RunPod & Massed Compute](https://furkangozukara.com/apps/wan2gp-1-click-installer-for-windows-runpod-massed-compute/): Wan2GP runs Wan 2.1 and Wan 2.2 video models like Wan 2.2 Animate with optimized VRAM use, and this automatic installer sets it up with uv about 100x faster. - [LTX 2 Ultimate Tutorial: ComfyUI & SwarmUI on Windows and Cloud](https://furkangozukara.com/apps/ltx-2-ultimate-tutorial-comfyui-swarmui-on-windows-and-cloud/): LTX2 video tutorial covering ComfyUI and SwarmUI setup, model downloads and ready presets plus Z-Image Base, on GPUs with as low as 6 GB VRAM. - [Wan 2.2 LoRA Training Tutorial: Text to Image & Video, Windows & Cloud](https://furkangozukara.com/apps/wan-2-2-lora-training-tutorial-text-to-image-video-windows-cloud/): Wan 2.2 complete LoRA training tutorial with Musubi Tuner, based on over 64 trainings; the ready configs train locally on GPUs with as little as 6 GB VRAM. - [FramePack 1-Click Installers: Image-to-Video on Just 6 GB VRAM](https://furkangozukara.com/apps/framepack-1-click-installers-image-to-video-on-just-6-gb-vram/): FramePack, lllyasviel's image-to-video model, makes 30 FPS videos up to 120 seconds long in an advanced Gradio app with Sage Attention and Teacache support. - [Ovi Tutorial: Open-Source Video With Synced Audio on 6 GB GPUs](https://furkangozukara.com/apps/ovi-tutorial-open-source-video-with-synced-audio-on-6-gb-gpus/): Ovi, the first-ever public open-source model to generate video and synchronized audio, is a local version of VEO 3 and SORA 2 that runs on 6 GB GPUs. - [Ovi Premium App: Text & Image to Video With Audio, 1-Click Install](https://furkangozukara.com/apps/ovi-premium-app-text-image-to-video-with-audio-1-click-install/): Ovi SECourses Premium app generates 121-frame videos with real audio from text or images, like VEO 3 and SORA 2, with block swapping for low VRAM GPUs. - [Ovi Released: Generate Videos With Audio Like VEO 3 or SORA 2 Locally](https://furkangozukara.com/apps/ovi-released-generate-videos-with-audio-like-veo-3-or-sora-2-locally/): Ovi is a free, open source model that generates videos with synchronized audio from text or images; our app runs it locally with as low as 8.2 GB VRAM. - [Wan 2.2, FLUX Krea & Qwen Image Upgraded: Fast 8-Step Presets Tutorial](https://furkangozukara.com/apps/wan-2-2-flux-krea-qwen-image-upgraded-fast-8-step-presets-tutorial/): Wan 2.2, Qwen Image, FLUX and FLUX Krea get new SwarmUI presets after a week of research, and new 8-step LoRAs make Wan 2.2 video and Qwen Image fast. - [Wan 2.2 & FLUX Krea Full Tutorial: 1-Click Install & SwarmUI Presets](https://furkangozukara.com/apps/wan-2-2-flux-krea-full-tutorial-1-click-install-swarmui-presets/): Wan 2.2 and FLUX Krea Dev in SwarmUI with a ComfyUI backend, using ready presets built from hundreds of parameter tests for videos from text or images. - [Wan Prompt Guide: How to Prompt Wan Models (Full Tutorial)](https://furkangozukara.com/apps/wan-prompt-guide-how-to-prompt-wan-models-full-tutorial/): Wan video prompting guide: the basic formula (subject, scene, motion), the advanced formula with aesthetic control and stylization, and prompt examples. - [MultiTalk Levelled Up: New ComfyUI Workflows for Better Animation](https://furkangozukara.com/apps/multitalk-levelled-up-new-comfyui-workflows-for-better-animation/): MultiTalk turns a static image and audio into talking and singing videos; 4 new ComfyUI workflow categories give better animation at the same VRAM and speed. - [V-Express Talking Avatar App: 1-Click Image to Animation Installers](https://furkangozukara.com/apps/v-express-talking-avatar-app-1-click-image-to-animation-installers/): Tencent AI Lab V-Express animates a still portrait into a talking video from audio; after VRAM optimizations 8 GB GPUs can make 20-second videos or longer. - [MultiTalk Super Charged: New Workflows, Better Animation, Same VRAM](https://furkangozukara.com/apps/multitalk-super-charged-new-workflows-better-animation-same-vram/): MultiTalk image-to-video workflows for ComfyUI, improved after over 1 day of tests on 8x A6000 GPUs: 4 workflow categories with better animation. - [EasyAnimate 1-Click Installer: Image-to-Video and Text-to-Video App](https://furkangozukara.com/apps/easyanimate-1-click-installer-image-to-video-and-text-to-video-app/): EasyAnimate 5.1 app for text-to-video, image-to-video with start and end frames and video-to-video, at 512, 768 and 960 resolutions; 512 needs a 24 GB GPU. - [Wan 2.1 FusionX & Self Forcing LoRA: 8-Step Video + FLUX Upscaling Guide](https://furkangozukara.com/apps/wan-2-1-fusionx-self-forcing-lora-8-step-video-flux-upscaling-guide/): Wan 2.1 FusionX and Self Forcing LoRA generate local videos from text or images in only 8 steps with our SwarmUI presets, plus FLUX 2x latent upscaling. - [SwarmUI Beginner's Guide: Generate AI Videos With Wan 2.1 and More](https://furkangozukara.com/apps/swarmui-beginner-s-guide-generate-ai-videos-with-wan-2-1-and-more/): SwarmUI beginner's guide to AI video by its developer mcmonkey: pick a model such as Wan 2.1 14B, add it to the right folder and generate your first clips. - [FramePack Studio 1-Click Installer for Windows, RunPod & Massed Compute](https://furkangozukara.com/apps/framepack-studio-1-click-installer-for-windows-runpod-massed-compute/): FramePack Studio, a popular fork of FramePack, installs in 1 click with pre-compiled libraries that support RTX 5000 series and older GPUs. - [Wan 2.1 Gradio App: 1-Click Install, 4 GB VRAM, T2V, I2V & V2V](https://furkangozukara.com/apps/wan-2-1-gradio-app-1-click-install-4-gb-vram-t2v-i2v-v2v/): Wan 2.1 Gradio app for text to video, image to video and video to video that runs with as low as 4 GB VRAM and batch processes whole folders. - [VEO 3 Flow Full Tutorial: How to Use VEO 3 in Google Flow](https://furkangozukara.com/apps/veo-3-flow-full-tutorial-how-to-use-veo-3-in-google-flow/): VEO 3 in Flow, Google's official tutorial, slightly edited: make cinematic video clips with native audio in the Flow AI filmmaking tool, with chapters. - [SwarmUI Teacache Tutorial: Wan 2.1 I2V & T2V Presets, 2x Faster](https://furkangozukara.com/apps/swarmui-teacache-tutorial-wan-2-1-i2v-t2v-presets-2x-faster/): SwarmUI Teacache full tutorial: speed up Wan 2.1 image-to-video and text-to-video 2x and more with ready presets, using ComfyUI as the backend. - [VEO 3 AI Video Generation With Native Audio: 60 User-Generated Examples](https://furkangozukara.com/apps/veo-3-ai-video-generation-with-native-audio-60-user-generated-examples/): Google VEO 3 generates short videos with native audio, dialogue and sound effects; a 4K video shows 60 wild user-generated examples from its launch. - [Wan 2.1 T2V & I2V in SwarmUI Tutorial: CausVid LoRA for Extreme Speed](https://furkangozukara.com/apps/wan-2-1-t2v-i2v-in-swarmui-tutorial-causvid-lora-for-extreme-speed/): Wan 2.1 text-to-video and image-to-video in SwarmUI with the CausVid LoRA at just 8 steps; on an RTX 5090 a clip took 90 seconds versus 170 on an RTX 3090 Ti. - [FantasyTalking App: 1-Click Realistic Talking Portrait Generation](https://furkangozukara.com/apps/fantasytalking-app-1-click-realistic-talking-portrait-generation/): FantasyTalking SECourses app turns a portrait and an audio clip into a realistic talking video, with VRAM presets, batch processing and multi-line prompts. - [FramePack 1-Click Installers Released: 6 GB VRAM Image-to-Video](https://furkangozukara.com/apps/framepack-1-click-installers-released-6-gb-vram-image-to-video/): FramePack image-to-video works with as low as 6 GB VRAM and can generate videos up to 120 seconds; longer videos keep the same speed and VRAM use. - [FramePack Full Tutorial: 120 Second Image-to-Video With 6 GB VRAM](https://furkangozukara.com/apps/framepack-full-tutorial-120-second-image-to-video-with-6-gb-vram/): FramePack by lllyasviel turns images into consistent videos up to 120 seconds on GPUs with as little as 6GB; the tutorial shows 1-click install on Windows. - [Wan 2.1 Gradio App Updates Tutorial: Squish Effect LoRA Videos Locally](https://furkangozukara.com/apps/wan-2-1-gradio-app-updates-tutorial-squish-effect-lora-videos-locally/): Wan 2.1 Gradio app now supports LoRAs; this tutorial makes squishing videos with the Squish Effect LoRA and shows RIFE FPS boost, FP8/BF16 and config saving. - [Wan 2.1 Gradio App Released: T2V, I2V and V2V From 4 GB VRAM](https://furkangozukara.com/apps/wan-2-1-gradio-app-released-t2v-i2v-and-v2v-from-4-gb-vram/): Wan 2.1 Ultra Advanced Gradio app release: text to video, image to video and video to video, with the 1.3B model running in as low as 3.5 GB VRAM. - [CogVideoX1.5-5B-I2V Image to Video: Low VRAM Test With MMAudio Sound](https://furkangozukara.com/apps/cogvideox1-5-5b-i2v-image-to-video-low-vram-test-with-mmaudio-sound/): CogVideoX1.5-5B-I2V is a decent open source image-to-video model optimized for low VRAM; native resolution is 1360px with up to 10 seconds and 161 frames. - [Kling AI Video Guide: Free Text-to-Video & Image-to-Video Worldwide](https://furkangozukara.com/apps/kling-ai-video-guide-free-text-to-video-image-to-video-worldwide/): Kling AI video is now released globally and free to use; this complete guide covers email sign-up and unfiltered text-to-video and image-to-video tests. - [LivePortrait Tutorial: Static Image to Video and Video to Video](https://furkangozukara.com/apps/liveportrait-tutorial-static-image-to-video-and-video-to-video/): LivePortrait is a state of the art AI model that turns a static photo and a driving video into a talking animation in seconds, with video to video support. ## AI Image Generation and Editing Apps - [MiniMax H3 Image: New Preset Fixes Tiny-Detail Errors at 16 MP](https://furkangozukara.com/apps/minimax-h3-image-new-preset-fixes-tiny-detail-errors-at-16-mp/): MiniMax H3 Image gets a new text-to-image preset that draws small objects such as watch gears, gem settings and tools more correctly at 16 MP in 6 steps. - [MiniMax H3 Image in ComfyUI: 16 MP Images and Edits in 6 Steps](https://furkangozukara.com/apps/minimax-h3-image-in-comfyui-16-mp-images-and-edits-in-6-steps/): MiniMax H3 Image is a new 33.1B model that generates and edits 16 MP pictures in 6 steps on your own GPU; our ComfyUI V150 ships two ready presets for it. - [SwarmUI Full Tutorial: 1-Click Install, 4 MP Images, HD Video & Lip Sync](https://furkangozukara.com/apps/swarmui-full-tutorial-1-click-install-4-mp-images-hd-video-lip-sync/): SwarmUI full tutorial from a fresh Windows install: our installers, model downloader with 517+ models and 86 presets turn your PC into a complete AI studio. - [ComfyUI Install & FLUX 3 Klein: The Complete Beginner Tutorial](https://furkangozukara.com/apps/comfyui-install-flux-3-klein-the-complete-beginner-tutorial/): ComfyUI beginner tutorial: install on Windows and learn FLUX 3 Klein with 16 core examples, precise image edits, RTX 5090 tests and launcher tuning. - [Ideogram 4 JSON Prompting Tutorial With Ultimate Image Captioner Pro](https://furkangozukara.com/apps/ideogram-4-json-prompting-tutorial-with-ultimate-image-captioner-pro/): Ideogram 4 runs locally in SwarmUI and ComfyUI with Turbo, Balanced and Highest Quality presets; the masterclass teaches accurate JSON prompts. - [Stable Diffusion WebUI Forge NEO: 1-Click Installer for Windows & Cloud](https://furkangozukara.com/apps/stable-diffusion-webui-forge-neo-1-click-installer-for-windows-cloud/): Forge NEO is the up-to-date, maintained Forge Web UI; the installer sets it up with Torch 2.13.0, CUDA 13 and precompiled xFormers and Flash Attention. - [Paints-UNDO Premium App: 1-Click Installers for Windows & Cloud](https://furkangozukara.com/apps/paints-undo-premium-app-1-click-installers-for-windows-cloud/): Paints-UNDO from lllyasviel turns any image into drawing steps and a video that recreates it from scratch, using 7 GB VRAM at 1024px with optimizations. - [Automatic1111 SD Web UI 1-Click Installer: SDXL, ControlNet, xFormers](https://furkangozukara.com/apps/automatic1111-sd-web-ui-1-click-installer-sdxl-controlnet-xformers/): Automatic1111 SD Web UI 1-click installer V13 for Windows, RunPod and Massed Compute with ADetailer, ControlNet, Deforum and xFormers, tested on an RTX 5090. - [NVFP4 Model Converter Gradio App for FLUX: Windows & Linux](https://furkangozukara.com/apps/nvfp4-model-converter-gradio-app-for-flux-windows-linux/): NVFP4 Model Converter app quantizes FLUX models to NVFP4, which runs 100%+ faster than BF16, FP8 or GGUF; the FLUX SRPO NVFP4 result is only 6.32 GB. - [Qwen Image Edit 2511 Tutorial: 12-Case Comparison With 2509](https://furkangozukara.com/apps/qwen-image-edit-2511-tutorial-12-case-comparison-with-2509/): Qwen Image Edit 2511 is a free, open source image editor that challenges Nano Banana Pro, with native 2560x2560 output in 12 steps, even on 6 GB GPUs. - [Z-Image Turbo Tutorial: LoRA Training, ControlNet Union & Presets](https://furkangozukara.com/apps/z-image-turbo-tutorial-lora-training-controlnet-union-presets/): Z-Image Turbo step by step: generate with ready SwarmUI presets, add Fun ControlNet Union and train LoRAs with Ostris AI Toolkit on GPUs from 6 GB. - [BiRefNet HR Background Remover: 1-Click Gradio App for Windows & Cloud](https://furkangozukara.com/apps/birefnet-hr-background-remover-1-click-gradio-app-for-windows-cloud/): BiRefNet HR removes backgrounds at high resolution in single or batch mode, with BF16, FP16 or FP32 precision and a No Resize mode that keeps native size. - [Qwen Image Next Level Realism: LoRA Workflow Tutorial, 4+4 Steps](https://furkangozukara.com/apps/qwen-image-next-level-realism-lora-workflow-tutorial-4-4-steps/): Qwen Image presets for SwarmUI with new realism LoRAs reach next level realism in only 4+4 steps; the full tutorial also covers inpainting and outpainting. - [SwarmUI and ComfyUI Tutorial: Free Local AI Image Generation](https://furkangozukara.com/apps/swarmui-and-comfyui-tutorial-free-local-ai-image-generation/): SwarmUI with a ComfyUI backend gives free, local AI image generation: a 1-click installer, realism and style presets, text-based editing and 4K/8K upscaling. - [Hunyuan Image 2.1 Full Tutorial: 1-Click Local & Cloud Install](https://furkangozukara.com/apps/hunyuan-image-2-1-full-tutorial-1-click-local-cloud-install/): Tencent Hunyuan Image 2.1 tutorial: install our Gradio app in 1 click and make 2560x1440 images with prompt auto-enhance and an 8-step refiner. - [Stable Diffusion WebUI Forge Classic: 1-Click Maintained Automatic1111](https://furkangozukara.com/apps/stable-diffusion-webui-forge-classic-1-click-maintained-automatic1111/): Forge Classic is the up-to-date, maintained Automatic1111 Web UI; its smart launcher adds GPU-specific optimization flags automatically. - [HunyuanImage 2.1 Gradio App: 1-Click Install for Windows & Cloud](https://furkangozukara.com/apps/hunyuanimage-2-1-gradio-app-1-click-install-for-windows-cloud/): Tencent HunyuanImage 2.1 text-to-image app with a refiner and image refinement tab; the distilled model makes an image in about 5 seconds on an RTX 6000 PRO. - [GenTube Review & Tutorial: Free AI Image Generation in 2 Seconds](https://furkangozukara.com/apps/gentube-review-tutorial-free-ai-image-generation-in-2-seconds/): GenTube is a new free image generation platform that makes AI art in about 2 seconds; the review covers live prompting, remixes, Magic Edit and channels. - [Nano Banana (Gemini 2.5 Flash Image) vs Qwen Image Edit: Full Tutorial](https://furkangozukara.com/apps/nano-banana-gemini-2-5-flash-image-vs-qwen-image-edit-full-tutorial/): Nano Banana image editing tested live in 27 unique cases against the free, local Qwen Image Edit, with every prompt, image and result shown in the tutorial. - [Qwen Image Edit Full Tutorial: 26 Demo Cases, Prompts & Images](https://furkangozukara.com/apps/qwen-image-edit-full-tutorial-26-demo-cases-prompts-images/): Qwen Image Edit full tutorial with 26 different demo cases, from photo restoration and text changes to stylization, object removal and outpainting in SwarmUI. - [Wan 2.2, FLUX Krea & Qwen Image Upgraded: Fast 8-Step Presets Tutorial](https://furkangozukara.com/apps/wan-2-2-flux-krea-qwen-image-upgraded-fast-8-step-presets-tutorial/): Wan 2.2, Qwen Image, FLUX and FLUX Krea get new SwarmUI presets after a week of research, and new 8-step LoRAs make Wan 2.2 video and Qwen Image fast. - [Qwen Image Showcase: Complex Prompts & Text Rendering for Thumbnails](https://furkangozukara.com/apps/qwen-image-showcase-complex-prompts-text-rendering-for-thumbnails/): Qwen Image, free and open source, is unchallenged at complex prompts and text on images; these raw SwarmUI thumbnails used Gemini 2.5 Pro prompts. - [Qwen Image vs FLUX Dev vs FLUX Krea: New Text-to-Image King Comparison](https://furkangozukara.com/apps/qwen-image-vs-flux-dev-vs-flux-krea-new-text-to-image-king-comparison/): Qwen Image compared with FLUX Dev and FLUX Krea over 700 generations; Qwen follows prompts far better and has a 68% bigger native resolution than FLUX Dev. - [Wan 2.2 & FLUX Krea Full Tutorial: 1-Click Install & SwarmUI Presets](https://furkangozukara.com/apps/wan-2-2-flux-krea-full-tutorial-1-click-install-swarmui-presets/): Wan 2.2 and FLUX Krea Dev in SwarmUI with a ComfyUI backend, using ready presets built from hundreds of parameter tests for videos from text or images. - [InstantID Gradio App: 1-Click Face Transfer With Xinsir ControlNet](https://furkangozukara.com/apps/instantid-gradio-app-1-click-face-transfer-with-xinsir-controlnet/): InstantID Gradio app transfers a face from one photo into new SDXL images with Xinsir ControlNet; block swap lowers 1280x1280 runs to 8.25 GB VRAM. - [NVIDIA SANA 1-Click Installers & Gradio App: 4K Images on 8 GB VRAM](https://furkangozukara.com/apps/nvidia-sana-1-click-installers-gradio-app-4k-images-on-8-gb-vram/): NVIDIA Labs SANA text-to-image model in an improved SECourses Gradio app; with all 4 VRAM optimizations, 1K images need 4 GB, 2K 6 GB and 4K 8 GB GPUs. - [FLUX Kontext Dev Tutorial: Local Image Editing on Windows](https://furkangozukara.com/apps/flux-kontext-dev-tutorial-local-image-editing-on-windows/): FLUX Kontext Dev edits any part of an image with just a prompt, no mask or ControlNet, and can restore, colorize, outpaint and combine images in SwarmUI. - [IDM-VTON Virtual Try-On App: Windows, Cloud & Free Kaggle Installers](https://furkangozukara.com/apps/idm-vton-virtual-try-on-app-windows-cloud-free-kaggle-installers/): IDM-VTON virtual try-on app that puts any clothing, and even objects, onto a person photo, with FP16 for the best results or 4-bit quantization for low VRAM. - [Tencent InstantCharacter App: 1-Click Install for Windows & Cloud](https://furkangozukara.com/apps/tencent-instantcharacter-app-1-click-install-for-windows-cloud/): Tencent InstantCharacter Gradio app, fixed and improved over the official repo, puts a reference character into new FLUX images using about 22 GB VRAM. - [ComfyUI Clothing Transfer & 3D Figure Workflows: 1-Click Tutorial](https://furkangozukara.com/apps/comfyui-clothing-transfer-3d-figure-workflows-1-click-tutorial/): ComfyUI workflows that transfer any clothing onto a new person and turn any person into a 3D toy figure, installed in 1 click and run with FLUX Dev models. - [IC-Light Relight Images Gradio App: 1-Click Windows, Cloud & Kaggle](https://furkangozukara.com/apps/ic-light-relight-images-gradio-app-1-click-windows-cloud-kaggle/): IC-Light from lllyasviel relights your images in an advanced Gradio app, with a second launcher that lets you set your own background. - [Tencent InstantCharacter Released: 1-Click Installers & Improved App](https://furkangozukara.com/apps/tencent-instantcharacter-released-1-click-installers-improved-app/): Tencent InstantCharacter improved app with a Character Scale setting and FLUX LoRA support; it fixes the broken official Gradio demo and saves every image. - [ByteDance UNO Gradio App: 1-Click Installers for Windows & Cloud](https://furkangozukara.com/apps/bytedance-uno-gradio-app-1-click-installers-for-windows-cloud/): ByteDance UNO improved Gradio app with 1-click installers; it downloads FLUX Dev and all needed models automatically and supports RTX 5000 series GPUs. - [HiDream NF4 Gradio App: 1-Click Installer for Windows & Cloud](https://furkangozukara.com/apps/hidream-nf4-gradio-app-1-click-installer-for-windows-cloud/): HiDream text-to-image in NF4 quantized form uses less VRAM than the original model; the app downloads Fast first, then Dev and Full only when you use them. - [MagicQuill Inpainting App: 1-Click Installers for Windows & Cloud](https://furkangozukara.com/apps/magicquill-inpainting-app-1-click-installers-for-windows-cloud/): MagicQuill interactive image editing system with an ultra fast inpaint model; the update adds RTX 5000 support, higher resolution and image auto-save. - [Invoke AI 1-Click Installers for Windows, RunPod & Massed Compute](https://furkangozukara.com/apps/invoke-ai-1-click-installers-for-windows-runpod-massed-compute/): Invoke AI image generation app in its latest version, with 1-click installers and a video tutorial; an extra upgrade file makes it run on RTX 5000 GPUs. - [IDM-VTON Virtual Try-On App Released: Installers & Free Kaggle Notebook](https://furkangozukara.com/apps/idm-vton-virtual-try-on-app-released-installers-free-kaggle-notebook/): IDM-VTON virtual try-on app dresses anyone in any clothing and can transfer objects too; it handles any resolution and aspect ratio with manual masking. - [BiRefNet HR Released: Batch Background Remover App for Windows & Cloud](https://furkangozukara.com/apps/birefnet-hr-released-batch-background-remover-app-for-windows-cloud/): BiRefNet HR brings a 2048x2048 high resolution background removal model to our upgraded batch processing Gradio app, with speed and VRAM optimizations. - [Paints-UNDO Installers Released: Undo Images Like Drawing From Scratch](https://furkangozukara.com/apps/paints-undo-installers-released-undo-images-like-drawing-from-scratch/): Paints-UNDO takes an input image and mimics how an artist could have drawn it, outputting drawing steps and a video, with a low VRAM mode that works great. - [SANA 4K: 16 MP (4096x4096) Images Under 8 GB VRAM With New Optimizations](https://furkangozukara.com/apps/sana-4k-16-mp-4096x4096-images-under-8-gb-vram-with-new-optimizations/): SANA 4K model now generates 16 megapixel (4096x4096) raw images with under 8 GB VRAM after the app moved to the Diffusers pipeline with big VRAM optimizations. - [IC-Light Image Relighting App Released: Windows, Cloud & Kaggle](https://furkangozukara.com/apps/ic-light-image-relighting-app-released-windows-cloud-kaggle/): The famous IC-Light relights images in a Gradio app more advanced than the official repo, with 1-click installers for Windows, RunPod and Massed Compute. - [SANA Tutorial: Fast Ultra HD Text to Image on Windows, Cloud & Kaggle](https://furkangozukara.com/apps/sana-tutorial-fast-ultra-hd-text-to-image-on-windows-cloud-kaggle/): NVIDIA SANA 2K is a fast 4 megapixel text to image model; this tutorial uses our Gradio app to generate 2048x2048 images on a PC and in the cloud. - [FLUX Tools SwarmUI Tutorial: Inpainting, Outpainting, Redux & Canny](https://furkangozukara.com/apps/flux-tools-swarmui-tutorial-inpainting-outpainting-redux-canny/): FLUX.1 Tools complete SwarmUI tutorial: inpaint and outpaint with FLUX Fill, restyle and combine images with Redux, and guide images with Depth and Canny. - [NVIDIA SANA Model Released: Speed and VRAM Tests, 1-Click Installers](https://furkangozukara.com/apps/nvidia-sana-model-released-speed-and-vram-tests-1-click-installers/): NVIDIA Labs developed SANA and published its weights and a Gradio demo app; with default settings an image takes 2 s on an RTX 4090 and 9.5 s on an RTX 3060. - [FLUX Models Speed & VRAM Comparison: Every Version and Precision](https://furkangozukara.com/apps/flux-models-speed-vram-comparison-every-version-and-precision/): FLUX Dev and Schnell in FP16, FP8 and NF4 compared for peak VRAM usage and speed in SwarmUI on an A6000; flux1-dev-fp8 uses 19.33 GB and takes 28 seconds. - [IP-Adapter-FaceID-PlusV2: 0-Shot Face Transfer Auto Installer & App](https://furkangozukara.com/apps/ip-adapter-faceid-plusv2-0-shot-face-transfer-auto-installer-app/): IP-Adapter-FaceID-PlusV2 Gradio app puts a face into new SD 1.5 and SDXL images with no training, and caches face embeddings for faster generation. - [Shopify Background Replacement App for Product Images, 1-Click](https://furkangozukara.com/apps/shopify-background-replacement-app-for-product-images-1-click/): Shopify background replacement app that replaces the background of anything, especially product images; best at 1024x1024 and blazing fast on an RTX 5090. ## Image and Video Upscaling and Restoration - [SECourses Upscaler Pro: SeedVR2 & FlashVSR+ Video and Image Upscaler](https://furkangozukara.com/apps/secourses-upscaler-pro-seedvr2-flashvsr-video-and-image-upscaler/): SECourses Upscaler Pro turns low-resolution videos and images into sharp 1080p, 2K and 4K results with nine AI engines in one standalone app. - [SECourses Upscaler Pro Tutorial: FlashVSR+ & SeedVR2.5 vs Topaz AI](https://furkangozukara.com/apps/secourses-upscaler-pro-tutorial-flashvsr-seedvr2-5-vs-topaz-ai/): SECourses Upscaler Pro beats Topaz AI by far in side-by-side video tests, using specialized FlashVSR+ and SeedVR2.5 on local GPUs with as little as 8 GB VRAM. - [SeedVR2 & FlashVSR+ Image and Video Upscaler Pro V1 Released](https://furkangozukara.com/apps/seedvr2-flashvsr-image-and-video-upscaler-pro-v1-released/): SECourses Ultimate Video and Image Upscaler Pro 1.0 release video: SeedVR2, FlashVSR+, GAN upscalers and RIFE frame interpolation with a full queue system. - [SUPIR 1-Click Installer for Windows, Linux, Cloud & Free Kaggle](https://furkangozukara.com/apps/supir-1-click-installer-for-windows-linux-cloud-free-kaggle/): SUPIR upscales and restores images better than Magnific, for example a 3x upscale from 1024 to 3072 px with face restore, and now supports up to 4 LoRAs. - [SUPIR Image Upscaler Update: RTX 1000 to RTX 5000 & Cloud GPU Support](https://furkangozukara.com/apps/supir-image-upscaler-update-rtx-1000-to-rtx-5000-cloud-gpu-support/): SUPIR is still the unchallenged image upscaler; the V91 update brings Torch 2.8, CUDA 12.9, a resumable hash-checked model downloader and faster installs. - [AuraSR GigaGAN 4x Upscaler App With Seams Fix & Batch Processing](https://furkangozukara.com/apps/aurasr-gigagan-4x-upscaler-app-with-seams-fix-batch-processing/): AuraSR GigaGAN always upscales images 4x with the same VRAM for any input size; it is fast and VRAM friendly, so it works on weaker PCs. - [Ultimate Video Processing App: Increase FPS of Videos in Batch](https://furkangozukara.com/apps/ultimate-video-processing-app-increase-fps-of-videos-in-batch/): Ultimate Video Processing App increases the frame rate (FPS) of your existing videos, including batch processing of many videos; version 1 runs on Windows. - [ImageMagick Upscaler for Images With Text and Graphs: Gradio App](https://furkangozukara.com/apps/imagemagick-upscaler-for-images-with-text-and-graphs-gradio-app/): ImageMagick Gradio app for deterministic upscaling of images with tiny text and graphs, preserving text that diffusion or GAN upscalers usually break. - [ImageMagick Upscaler Gradio App Released for Text and Graph Images](https://furkangozukara.com/apps/imagemagick-upscaler-gradio-app-released-for-text-and-graph-images/): ImageMagick based upscaler app for images with tiny text and graphs, where diffusion or GAN upscalers break the text; it runs multi-threaded on the CPU. - [AuraSR GigaGAN 4x Upscaler Released: Fast Batch Upscaling Under 5 GB](https://furkangozukara.com/apps/aurasr-gigagan-4x-upscaler-released-fast-batch-upscaling-under-5-gb/): AuraSR GigaGAN 4x upscaler Gradio app by SECourses upscales real-world images to 4x resolution with batch folder processing and needs less than 5 GB VRAM. - [FLUX LoRA Upscale Comparison: Raw vs SwarmUI 2x vs SUPIR App](https://furkangozukara.com/apps/flux-lora-upscale-comparison-raw-vs-swarmui-2x-vs-supir-app/): FLUX LoRA output upscale comparison on an imgsli slider: the raw image vs the SwarmUI quick 2x upscale button vs our SUPIR app with face upscale on. - [SUPIR Online by SupPixel AI: Full Tutorial and Settings Tests](https://furkangozukara.com/apps/supir-online-by-suppixel-ai-full-tutorial-and-settings-tests/): SUPIR Online from the original SUPIR authors at SupPixel AI, tested on a real photo and a creature drawing with all settings compared in full resolution. - [Image Folders Merger Script: Merge Images Side by Side for Comparison](https://furkangozukara.com/apps/image-folders-merger-script-merge-images-side-by-side-for-comparison/): Image Folders Merger is a small Python script that merges images from two folders side by side, made to compare upscaling effects such as AuraSR GigaGAN. - [CodeFormer Face Restoration: 1-Click Installers for Windows & Cloud](https://furkangozukara.com/apps/codeformer-face-restoration-1-click-installers-for-windows-cloud/): CodeFormer robust face restoration and enhancement network as a Gradio app with batch folder processing that never overwrites existing images. - [SUPIR Comparison Video: Juggernaut-XL v9 Better With Lower Text Guidance](https://furkangozukara.com/apps/supir-comparison-video-juggernaut-xl-v9-better-with-lower-text/): SUPIR gets better the more you use it: a comparison video made automatically in Gradio shows Juggernaut-XL_v9 works better with a lower Text Guidance Scale. - [SUPIR Full Tutorial: 1-Click Install, Batch Upscale & Magnific Test](https://furkangozukara.com/apps/supir-full-tutorial-1-click-install-batch-upscale-magnific-test/): SUPIR image upscaler full 33-minute tutorial: install with 1 click on Windows or RunPod and Linux, run on 12 GB VRAM GPUs and batch upscale your images. ## Face Swap, Deepfake and Talking Avatar Apps - [Henge 1-Click Video Studio: Character Swap, Motion Transfer, Head Swap](https://furkangozukara.com/apps/henge-1-click-video-studio-character-swap-motion-transfer-head-swap/): Henge uses MiniMax H3 to swap characters, transfer motion, swap heads and sharpen videos; a 12 second Character Swap took about 4 minutes on an RTX 5090. - [VisoMaster Fusion 1-Click Installer: 0-Shot Face Swap & Deep Fake App](https://furkangozukara.com/apps/visomaster-fusion-1-click-installer-0-shot-face-swap-deep-fake-app/): VisoMaster Fusion is a state of the art face swap studio for images, video, webcam and VR180 footage, now with DeepFaceLab DFM model support and TensorRT. - [FaceFusion 3.5 1-Click Installers: DeepFake App for Windows & Cloud](https://furkangozukara.com/apps/facefusion-3-5-1-click-installers-deepfake-app-for-windows-cloud/): FaceFusion 3.5.2 deepfake and face swap app with 1-click installers that now use uv, so installs run up to 100x faster on RunPod and faster on Windows. - [Deep Live Cam 1-Click Windows Installer: Live Webcam Face Swap](https://furkangozukara.com/apps/deep-live-cam-1-click-windows-installer-live-webcam-face-swap/): Deep Live Cam, the viral real-time webcam face swap app, installs with 1 click on Windows; v3 moved to Torch 2.8, CUDA 12.9 and uv for a faster setup. - [Roop Unleashed 1-Click Installers: Easy Face Swap for Windows & Cloud](https://furkangozukara.com/apps/roop-unleashed-1-click-installers-easy-face-swap-for-windows-cloud/): Roop Unleashed is one of the best face swap and deep fake apps; these installers run it on Windows, RunPod and Massed Compute with RTX 5000 series support. - [InstantID Gradio App: 1-Click Face Transfer With Xinsir ControlNet](https://furkangozukara.com/apps/instantid-gradio-app-1-click-face-transfer-with-xinsir-controlnet/): InstantID Gradio app transfers a face from one photo into new SDXL images with Xinsir ControlNet; block swap lowers 1280x1280 runs to 8.25 GB VRAM. - [V-Express Talking Avatar App: 1-Click Image to Animation Installers](https://furkangozukara.com/apps/v-express-talking-avatar-app-1-click-image-to-animation-installers/): Tencent AI Lab V-Express animates a still portrait into a talking video from audio; after VRAM optimizations 8 GB GPUs can make 20-second videos or longer. - [MultiTalk Super Charged: New Workflows, Better Animation, Same VRAM](https://furkangozukara.com/apps/multitalk-super-charged-new-workflows-better-animation-same-vram/): MultiTalk image-to-video workflows for ComfyUI, improved after over 1 day of tests on 8x A6000 GPUs: 4 workflow categories with better animation. - [FantasyTalking App: 1-Click Realistic Talking Portrait Generation](https://furkangozukara.com/apps/fantasytalking-app-1-click-realistic-talking-portrait-generation/): FantasyTalking SECourses app turns a portrait and an audio clip into a realistic talking video, with VRAM presets, batch processing and multi-line prompts. - [Deep Live Cam Windows Installer Released: Live Webcam Face Swap](https://furkangozukara.com/apps/deep-live-cam-windows-installer-released-live-webcam-face-swap/): Deep Live Cam, the viral live web cam face swap app, now has an automatic Windows installer that uses less VRAM, runs faster and supports RTX 5000 GPUs. - [LivePortrait Human & Animal 1-Click Installers for Windows & Cloud](https://furkangozukara.com/apps/liveportrait-human-animal-1-click-installers-for-windows-cloud/): LivePortrait human and animal versions in an improved Gradio app; V22 (May 2026) fixes the installers and runs on all GPUs including RTX 5090. - [Rope Face Swap App: 1-Click Rope Live, Alucard & NEXT Installers](https://furkangozukara.com/apps/rope-face-swap-app-1-click-rope-live-alucard-next-installers/): Rope is an advanced open source deepfake and face swap app; these 1-click installers set up Rope Live and Rope Alucard with RTX 5000 GPU support. - [VisoMaster Face Swap Tutorial: 1-Click Install on Windows & Cloud](https://furkangozukara.com/apps/visomaster-face-swap-tutorial-1-click-install-on-windows-cloud/): VisoMaster is a new 0-shot face swap and deepfake app for images and videos with GPU acceleration; the screenshots show how to install and use it. - [LivePortrait Newest Version Released With a More Advanced Gradio App](https://furkangozukara.com/apps/liveportrait-newest-version-released-with-a-more-advanced-gradio-app/): LivePortrait newest version auto installers for Windows, RunPod, Massed Compute and a free Kaggle notebook, now with the LivePortrait Animal v1.1 model. - [Roop Unleashed Installers Released: Easy Face Swap for Windows & Cloud](https://furkangozukara.com/apps/roop-unleashed-installers-released-easy-face-swap-for-windows-cloud/): Roop Unleashed, one of the best easy face swap and deepfake apps, now has 1-click installer scripts in a separate post, with live webcam face replacement. - [LivePortrait Tutorial: Static Image to Video and Video to Video](https://furkangozukara.com/apps/liveportrait-tutorial-static-image-to-video-and-video-to-video/): LivePortrait is a state of the art AI model that turns a static photo and a driving video into a talking animation in seconds, with video to video support. - [CodeFormer Face Restoration: 1-Click Installers for Windows & Cloud](https://furkangozukara.com/apps/codeformer-face-restoration-1-click-installers-for-windows-cloud/): CodeFormer robust face restoration and enhancement network as a Gradio app with batch folder processing that never overwrites existing images. - [Rope Pearl Deepfake Tutorials: Full HD Face Swap on Windows & Cloud](https://furkangozukara.com/apps/rope-pearl-deepfake-tutorials-full-hd-face-swap-on-windows-cloud/): Rope, a 1-click open source deepfake app, swaps multiple faces in videos up to 4K; the tutorials cover Windows and a no-GPU Massed Compute cloud machine. - [IP-Adapter-FaceID-PlusV2: 0-Shot Face Transfer Auto Installer & App](https://furkangozukara.com/apps/ip-adapter-faceid-plusv2-0-shot-face-transfer-auto-installer-app/): IP-Adapter-FaceID-PlusV2 Gradio app puts a face into new SD 1.5 and SDXL images with no training, and caches face embeddings for faster generation. ## Voice, Text-to-Speech, Transcription and Music AI - [ACE-Step 1.5 XL Premium: Music & Song Generator With 1-Click Install](https://furkangozukara.com/apps/ace-step-1-5-xl-premium-music-song-generator-with-1-click-install/): ACE-Step 1.5 XL Premium generates full songs locally with the Turbo, SFT and Base models; the V8 INT8 update makes SFT and Base songs up to 1.9x faster. - [Ultimate TTS Tutorial: Voice Training & Cloning in 600+ Languages](https://furkangozukara.com/apps/ultimate-tts-tutorial-voice-training-cloning-in-600-languages/): Ultimate TTS full Windows tutorial covering voice cloning, automatic take selection, multilingual speech and OmniVoice full fine-tuning in 600+ languages. - [Ultimate TTS: Voice Cloning & Training App With IndexTTS 2.5 & OmniVoice](https://furkangozukara.com/apps/ultimate-tts-voice-cloning-training-app-with-indextts-2-5-omnivoice/): Ultimate Text to Speech Generator clones any voice from a short clip and trains your own voice; IndexTTS 2.5 made 31 s of speech in 18 s on an RTX 5090. - [Whisper-WebUI Premium Guide: Transcribe 10 Minutes in 4 Seconds](https://furkangozukara.com/apps/whisper-webui-premium-guide-transcribe-10-minutes-in-4-seconds/): Whisper-WebUI Premium turns video and audio into subtitles and transcripts on Windows; this full guide covers 3 engines, 6 export formats and batch folders. - [Whisper-WebUI Premium: Fast Speech to Text Transcription App, 1-Click](https://furkangozukara.com/apps/whisper-webui-premium-fast-speech-to-text-transcription-app-1-click/): Whisper-WebUI Premium turns video or audio into subtitles in 100 languages; its INT8 engine transcribed a 1 hour 29 minute lecture in 32 seconds. - [SECourses Video Captioner Pro: 1-Click Video, Image & Audio Captions](https://furkangozukara.com/apps/secourses-video-captioner-pro-1-click-video-image-audio-captions/): SECourses Video Captioner Pro captions video, images and sound, transcribes speech with Whisper and builds training datasets using 21 checkpoint variants. - [IndexTTS 2.5 Premium Guide: Voice Cloning, Training & Audiobooks](https://furkangozukara.com/apps/indextts-2-5-premium-guide-voice-cloning-training-audiobooks/): IndexTTS 2.5 Premium guide to the SECourses voice cloning and generation app: long-form narration, LoRA/DoRA voice training, batches and low-VRAM setup. - [ACE-Step 1.5 Remix, Paints-Undo & Whisper Premium: Local AI Update](https://furkangozukara.com/apps/ace-step-1-5-remix-paints-undo-whisper-premium-local-ai-update/): ACE-Step XL 1.5 Premium remix tutorial: change lyrics and language, tune strength and melody retention, regenerate parts; a full remix takes about 33 seconds. - [ACE-Step 1.5 XL Tutorial: Local Music Studio, SAM Audio & Auto-Editor](https://furkangozukara.com/apps/ace-step-1-5-xl-tutorial-local-music-studio-sam-audio-auto-editor/): ACE-Step XL 1.5 Premium full tutorial: generate, remix and repaint songs locally, split stems with SAM Audio and install on Windows or cloud GPUs. - [Canary Qwen 2.5B vs Whisper: Open Source Subtitle Generator Full Guide](https://furkangozukara.com/apps/canary-qwen-2-5b-vs-whisper-open-source-subtitle-generator-full-guide/): Canary Qwen 2.5B in the Whisper-WebUI Premium app reached 5.91% WER and up to 46x real-time speed, ahead of Whisper Large V3 for English subtitles. - [RVC WebUI: The Ultimate AI Voice Cloning Guide (Zero to Hero)](https://furkangozukara.com/apps/rvc-webui-the-ultimate-ai-voice-cloning-guide-zero-to-hero/): RVC WebUI tutorial for the SECourses Premium voice changer: convert speech or song vocals, change your voice live and train custom voices with under 4 GB VRAM. - [MMAudio 1-Click Installers: Generate Audio for Any Video](https://furkangozukara.com/apps/mmaudio-1-click-installers-generate-audio-for-any-video/): MMAudio generates matching sound for any video, amazing for AI generated videos and games; V12 uses CUDA 12.9, Torch 2.8 and Flash Attention. - [RVC WebUI Premium: 1-Click Voice Conversion and Training App](https://furkangozukara.com/apps/rvc-webui-premium-1-click-voice-conversion-and-training-app/): RVC WebUI Premium converts and trains voices, with real-time microphone voice changing, voice merging and batch folder processing, using under 4 GB VRAM. - [LeVo Song Generator: 1-Click Local Music App With Batch Features](https://furkangozukara.com/apps/levo-song-generator-1-click-local-music-app-with-batch-features/): SECourses LeVo Song Generator makes full songs locally at almost Suno AI Music level; with Int8 a 180-second song needs as low as 12 GB VRAM. - [Claude AI Edits Word & Excel Files, Plus Wan Training Arrives](https://furkangozukara.com/apps/claude-ai-edits-word-excel-files-plus-wan-training-arrives/): Claude AI now edits Word documents and Excel spreadsheets; the video also shows Wan and Qwen Image training in Musubi Tuner and IndexTTS2 voice cloning. - [MMAudio Tutorial: Open Source AI Audio Generator for Videos & Games](https://furkangozukara.com/apps/mmaudio-tutorial-open-source-ai-audio-generator-for-videos-games/): MMAudio makes synced sound for videos, images and text prompts; this tutorial installs it on Windows in 1 click and compares RTX 5090 and 3090 Ti speed. ## Image to 3D Generation Apps - [Hi3DGen Image to 3D Mesh Generator App: 1-Click Windows & Cloud](https://furkangozukara.com/apps/hi3dgen-image-to-3d-mesh-generator-app-1-click-windows-cloud/): Hi3DGen turns a single image into a detailed, UV-unwrapped 3D mesh on your own PC; on an RTX 5090 our dwarf example became a 344,622-face mesh in 1 minute. - [TRELLIS Image to 3D: 1-Click Installers With Multi-Image Input](https://furkangozukara.com/apps/trellis-image-to-3d-1-click-installers-with-multi-image-input/): TRELLIS V10 turns one image or 2 to 4 views of an object into a textured GLB mesh, a Gaussian splat PLY and an MP4 turntable preview, with low VRAM presets. - [TRELLIS 2 App: Image to 3D Mesh & Texture With UniRig Auto Rigging](https://furkangozukara.com/apps/trellis-2-app-image-to-3d-mesh-texture-with-unirig-auto-rigging/): Microsoft TRELLIS 2 premium app generates top quality 3D models with mesh and texture from images, with libraries precompiled for RTX 1000 to 5000 series GPUs. - [SAM 3D Objects Gradio App: 1-Click Image to 3D for Windows & Cloud](https://furkangozukara.com/apps/sam-3d-objects-gradio-app-1-click-image-to-3d-for-windows-cloud/): Meta SAM 3D Objects Gradio app reconstructs full 3D shape, texture and layout from one image; a 30 MP image takes under 1 minute on RTX 5090. - [TripoSR Premium Gradio App: Fast Image to 3D With 1-Click Installers](https://furkangozukara.com/apps/triposr-premium-gradio-app-fast-image-to-3d-with-1-click-installers/): TripoSR premium app makes fast feedforward 3D reconstruction from a single image, significantly improved over the official VAST-AI-Research repo app. - [Hi3DGen Full Tutorial: Image to 3D Mesh Generation on Windows & Cloud](https://furkangozukara.com/apps/hi3dgen-full-tutorial-image-to-3d-mesh-generation-on-windows-cloud/): Hi3DGen generates detailed 3D meshes from static images, better than TRELLIS and Hunyuan3D-2.0; the tutorial shows install, presets and batch processing. - [TRELLIS Image to 3D Tutorial: Multi-Angle Input, Works on 6 GB GPUs](https://furkangozukara.com/apps/trellis-image-to-3d-tutorial-multi-angle-input-works-on-6-gb-gpus/): TRELLIS is still the leading open source AI model for high-quality 3D assets from static images; this full tutorial shows our 1-click app and examples. - [VAST-AI TripoSF 3D Shape Gradio App: 1-Click Windows & Cloud Installer](https://furkangozukara.com/apps/vast-ai-triposf-3d-shape-gradio-app-1-click-windows-cloud-installer/): VAST-AI TripoSF Gradio app for 3D shapes downloads its model automatically; V6 adds batch processing and improved, fixed output naming. - [TRELLIS Image to 3D App Installers Updated: Multi-Image Support](https://furkangozukara.com/apps/trellis-image-to-3d-app-installers-updated-multi-image-support/): TRELLIS from Microsoft makes high quality GLB and Gaussian Splatting PLY files from images; the updated Windows installer ships pre-compiled libraries. ## Captioning, Dataset and Image Processing Tools - [SECourses Video Captioner Pro: 1-Click Video, Image & Audio Captions](https://furkangozukara.com/apps/secourses-video-captioner-pro-1-click-video-image-audio-captions/): SECourses Video Captioner Pro captions video, images and sound, transcribes speech with Whisper and builds training datasets using 21 checkpoint variants. - [Ultimate Image Processing App: 1-Click Batch Cropping, SAM 3.1 Masking](https://furkangozukara.com/apps/ultimate-image-processing-app-1-click-batch-cropping-sam-3-1-masking/): Ultimate Image Processing Premium turns raw photo folders into training-ready datasets with nine tools, from SAM 3.1 cropping to duplicate removal. - [Ultimate Image Processing App Released: Batch Crop, Resize & SAM 3.1](https://furkangozukara.com/apps/ultimate-image-processing-app-released-batch-crop-resize-sam-3-1/): Ultimate Image Processing Premium by SECourses turns a folder of raw photos into a clean, framed training dataset with nine tools, including SAM 3.1 cropping. - [Ultimate Image Captioner Pro: Qwen3 VL & JoyCaption JSON Captioning](https://furkangozukara.com/apps/ultimate-image-captioner-pro-qwen3-vl-joycaption-json-captioning/): Ultimate Image Captioner Pro writes JSON and text captions with Qwen3 VL 8B Instruct and Joy Caption models for Ideogram 4 JSON prompting and training. - [DeepFace Face Similarity Sorter: 1-Click Batch Image Sorting App](https://furkangozukara.com/apps/deepface-face-similarity-sorter-1-click-batch-image-sorting-app/): DeepFace face similarity sorter ranks folders of AI-generated images by how close each face is to your source photos and renames files by rank and score. - [All JoyCaption Apps: 1-Click Installers With Batch Captioning](https://furkangozukara.com/apps/all-joycaption-apps-1-click-installers-with-batch-captioning/): JoyCaption in four improved Gradio apps: Pre-Alpha, Alpha One, Alpha Two and the newest, fastest Joy Caption Beta One, all with batch and multi-GPU captioning. - [Video and Image Comparison Slider Apps With Batch Upload and Zoom](https://furkangozukara.com/apps/video-and-image-comparison-slider-apps-with-batch-upload-and-zoom/): Comparison slider apps for videos and images with batch upload and live change; extremely lightweight, run locally and need only Git to install. - [SECourses Caption Editor Pro: Easy Batch Caption Editing Gradio App](https://furkangozukara.com/apps/secourses-caption-editor-pro-easy-batch-caption-editing-gradio-app/): SECourses Caption Editor Pro edits image captions for training datasets one by one or in batch, with tag analytics, a token counter and a check-word tab. - [Image Captioning Editor Gradio App Released: Batch Caption Editing](https://furkangozukara.com/apps/image-captioning-editor-gradio-app-released-batch-caption-editing/): Image Captioning Editor by SECourses lets you browse image datasets, edit .txt captions one by one or in batch and count words and n-grams. - [CogVLM 2 Vision Model App: 1-Click Batch Image Captioning Installers](https://furkangozukara.com/apps/cogvlm-2-vision-model-app-1-click-batch-image-captioning-installers/): CogVLM 2, the most powerful open source vision model at GPT-4 level for image tasks, in a SECourses batch captioning app that runs fast in 4-bit mode. - [DeepFace Image Similarity Sorter App Released: Batch Sort AI Images](https://furkangozukara.com/apps/deepface-image-similarity-sorter-app-released-batch-sort-ai-images/): DeepFace image similarity sorter ranks a folder of AI or real images by face resemblance to your source photos and renames them by rank and distance score. - [CogVLM 2 Vision Model Gradio App Released: Caption Images and More](https://furkangozukara.com/apps/cogvlm-2-vision-model-gradio-app-released-caption-images-and-more/): CogVLM 2 is a powerful, free and open source vision model at GPT-4 level; our Gradio app captions images fast with 4-bit quantization. - [Perfect Quality Training Image Dataset for FLUX, SD 3.5, SDXL & SD 1.5](https://furkangozukara.com/apps/perfect-quality-training-image-dataset-for-flux-sd-3-5-sdxl-sd-1-5/): Example model training dataset of 28 maximum-quality 1024x1024 images, prepared to learn and test training, sized for the FLUX batch size 7 config. - [JoyCaption Gradio App for Training Data: Batch Captioning Overview](https://furkangozukara.com/apps/joycaption-gradio-app-for-training-data-batch-captioning-overview/): JoyCaption captions training images with an advanced Gradio app and 1-click install; the app supports Pre Alpha, Alpha One, Alpha Two and the fast Beta One. - [Image Folders Merger Script: Merge Images Side by Side for Comparison](https://furkangozukara.com/apps/image-folders-merger-script-merge-images-side-by-side-for-comparison/): Image Folders Merger is a small Python script that merges images from two folders side by side, made to compare upscaling effects such as AuraSR GigaGAN. - [Stable Diffusion Regularization Images: 4K Man & Woman Class Dataset](https://furkangozukara.com/apps/stable-diffusion-regularization-images-4k-man-woman-class-dataset/): Stable Diffusion regularization images: 5200 hand-checked real man and 5200 woman photos, cropped and sorted by face quality for SDXL and SD training. - [Convert Very Long X/Y/Z Plot Output Images Into Square Grids](https://furkangozukara.com/apps/convert-very-long-x-y-z-plot-output-images-into-square-grids/): X/Y/Z plot square grid script: cuts a very long comparison output image into chunks and arranges them on one square canvas, needing only Pillow. - [Image Validator Script for Training Datasets: Move Corrupted Images](https://furkangozukara.com/apps/image-validator-script-for-training-datasets-move-corrupted-images/): Image validator script that scans a training dataset folder and its subfolders, logs invalid images and moves corrupted ones that would break training. ## Cloud GPU Guides: RunPod, Massed Compute, SimplePod, Kaggle - [Hugging Face Upload & Download Notebook: Private Repos & Multi Commit](https://furkangozukara.com/apps/hugging-face-upload-download-notebook-private-repos-multi-commit/): Hugging Face notebook that uploads and downloads models and files fast, with private repositories and multi commit support; V23 handles more token cases. - [SimplePod AI: Cheap Cloud GPU Alternative to RunPod and Vast AI](https://furkangozukara.com/apps/simplepod-ai-cheap-cloud-gpu-alternative-to-runpod-and-vast-ai/): SimplePod AI is a cheap cloud GPU service with a ready SECourses template, so all our RunPod scripts work unchanged, with permanent disks faster than RunPod. - [ComfyUI & SwarmUI on Massed Compute and RunPod: Cloud GPU Tutorial](https://furkangozukara.com/apps/comfyui-swarmui-on-massed-compute-and-runpod-cloud-gpu-tutorial/): ComfyUI and SwarmUI set up in 1 click on Massed Compute and RunPod cloud GPUs, with multi-GPU backends for faster Qwen, FLUX and Wan 2.2 generation. - [Massed Compute SECourses Coupon Now Works on All GPUs](https://furkangozukara.com/apps/massed-compute-secourses-coupon-now-works-on-all-gpus/): Massed Compute SECourses coupon now works on every GPU, including the newly arrived 96 GB RTX 6000 PRO, which costs 41.56% more on RunPod. - [ComfyUI & SwarmUI on RunPod Tutorial: 1-Click Setup With RTX 5000 GPUs](https://furkangozukara.com/apps/comfyui-swarmui-on-runpod-tutorial-1-click-setup-with-rtx-5000-gpus/): ComfyUI and SwarmUI on RunPod, step by step: install both with 1 click, connect ComfyUI as the SwarmUI backend and generate FLUX images and Wan 2.1 videos. - [SwarmUI Free Kaggle Notebook: FLUX, SDXL and SD 3.5 on Dual T4 GPUs](https://furkangozukara.com/apps/swarmui-free-kaggle-notebook-flux-sdxl-and-sd-3-5-on-dual-t4-gpus/): SwarmUI notebook for a free Kaggle account that runs FLUX Dev Q8 GGUF on both T4 GPUs at once, generating 2 images at the same time, with a video tutorial. - [MimicPC Tutorial: Easy Cloud Platform to Run AI Apps Online](https://furkangozukara.com/apps/mimicpc-tutorial-easy-cloud-platform-to-run-ai-apps-online/): MimicPC is a cloud platform and online AI generator that runs apps like ComfyUI with FLUX, CogVideoX and LivePortrait in your browser with permanent storage. - [Massed Compute vs RunPod: Why I Prefer Massed Compute for AI Apps](https://furkangozukara.com/apps/massed-compute-vs-runpod-why-i-prefer-massed-compute-for-ai-apps/): Massed Compute vs RunPod: of 3 identical RTX 4090 RunPod pods used to install SUPIR, only 1 worked, while Massed Compute stays fast with no errors. ## PC Setup, CUDA and Troubleshooting - [Windows & Ubuntu AI Requirements Installer: Full 1-Click Tutorial](https://furkangozukara.com/apps/windows-ubuntu-ai-requirements-installer-full-1-click-tutorial/): AI Requirements Installers take a Windows or Ubuntu PC from zero to AI-ready in one click: NVIDIA driver, CUDA 12.9 and 13.2, cuDNN 9.20 and Python 3.10-3.14. - [AI Requirements Installers: 1-Click Windows & Ubuntu PC Setup](https://furkangozukara.com/apps/ai-requirements-installers-1-click-windows-ubuntu-pc-setup/): AI Requirements Installers make a Windows or Ubuntu PC ready for local AI apps: NVIDIA driver, Python 3.10 to 3.14, CUDA 13.2, cuDNN 9.20 and C++ tools. - [Hugging Face Upload & Download Notebook: Private Repos & Multi Commit](https://furkangozukara.com/apps/hugging-face-upload-download-notebook-private-repos-multi-commit/): Hugging Face notebook that uploads and downloads models and files fast, with private repositories and multi commit support; V23 handles more token cases. - [CUDA and C++ Tools Compiling Errors: Windows Fix Guide for AI Apps](https://furkangozukara.com/apps/cuda-and-c-tools-compiling-errors-windows-fix-guide-for-ai-apps/): CUDA and C++ build errors in Python AI apps are solved once and for all: set up Visual Studio, CUDA 13.1 and 12.9.1, cuDNN, FFmpeg and TensorRT on Windows. - [Pre-Compiled Flash Attention & xFormers Wheels + CUDA Archs Guide](https://furkangozukara.com/apps/pre-compiled-flash-attention-xformers-wheels-cuda-archs-guide/): Pre-compiled AI library wheels for Windows and Linux, such as Flash Attention 2.8.4 for Torch 2.11 and CUDA 13, built with the right CUDA archs for each GPU. - [Hugging Face Invalid Credentials in Authorization Header Error: Fix](https://furkangozukara.com/apps/hugging-face-invalid-credentials-in-authorization-header-error-fix/): Hugging Face 401 Invalid credentials in Authorization header error appears when your saved token is no longer authorized; huggingface-cli logout fixes it. - [System Resource Monitor: Track RAM & VRAM Usage of AI Apps Live](https://furkangozukara.com/apps/system-resource-monitor-track-ram-vram-usage-of-ai-apps-live/): System Resource Monitor is a simple Python app that shows live RAM and VRAM usage and logs it to a CSV file, useful for tracking AI apps. - [How to Get Installation Logs and Change the Hugging Face Cache Folder](https://furkangozukara.com/apps/how-to-get-installation-logs-and-change-the-hugging-face-cache-folder/): Installation logs and the Hugging Face cache folder explained: how to get install logs to report errors, and how to move where AI models are auto-downloaded. - [Clonezilla Tutorial: Clone a Boot Disk to a Larger Disk for Free](https://furkangozukara.com/apps/clonezilla-tutorial-clone-a-boot-disk-to-a-larger-disk-for-free/): Clonezilla full tutorial: clone a smaller boot disk onto a larger SSD for free, keeping boot and partition data, then verify every cloned file bit by bit. - [Bit-by-Bit Disk & File Verification Software in C# With Source Code](https://furkangozukara.com/apps/bit-by-bit-disk-file-verification-software-in-c-with-source-code/): Bit-by-bit disk and file verification software in C#, fully multi-threaded, to verify a disk clone; full source code and an x64 EXE are included. ## Research, Comparisons and Experiments - [X Algorithm Update: Deep Dive Into the Latest x-algorithm Commit](https://furkangozukara.com/apps/x-algorithm-update-deep-dive-into-the-latest-x-algorithm-commit/): X's For You recommendation algorithm got a bigger update than it looks: commit e414c17 changed 187 files with 18,263 additions across the feed stack. - [BF16 vs GGUF Q8 vs FP8 Scaled vs NVFP4: Quality and Speed Comparison](https://furkangozukara.com/apps/bf16-vs-gguf-q8-vs-fp8-scaled-vs-nvfp4-quality-and-speed-comparison/): BF16, GGUF Q8, FP8 Scaled and NVFP4 precisions compared for speed and quality in ComfyUI with CUDA 13; NVFP4 ran 87% faster than GGUF Q8 on Z Image Turbo. - [NVFP4 With CUDA 13 Tutorial: 100%+ Speed Gain & Quality Comparison](https://furkangozukara.com/apps/nvfp4-with-cuda-13-tutorial-100-speed-gain-quality-comparison/): NVFP4 models run over 100% faster in ComfyUI and SwarmUI with CUDA 13; this full tutorial compares NVFP4 quality on FLUX 2, Z Image Turbo and FLUX 1 in grids. - [Qwen Image 2512 Training Comparison: The Best Model to Train](https://furkangozukara.com/apps/qwen-image-2512-training-comparison-the-best-model-to-train/): Qwen Image 2512 is the new king for training: the video compares trained Base, 2512, Edit 2509 and Edit 2511 models and shows a big leap in realism. - [Qwen Image 2512 First Training: 120 vs 180 vs 240 Epochs Comparison](https://furkangozukara.com/apps/qwen-image-2512-first-training-120-vs-180-vs-240-epochs-comparison/): Qwen Image 2512 first training results: the model behaves like FLUX Dev, and more training makes it more realistic with less noise on a 28-image dataset. - [FLUX 2 vs FLUX SRPO & New Kohya SS GUI Premium FLUX Training App](https://furkangozukara.com/apps/flux-2-vs-flux-srpo-new-kohya-ss-gui-premium-flux-training-app/): FLUX 2 Pro compared with a trained FLUX SRPO model, plus the new Kohya trainer V35 that trains FLUX on GPUs with as little as 6 GB VRAM. - [Qwen Image vs FLUX SRPO Training Comparison: 20 Images, Same Dataset](https://furkangozukara.com/apps/qwen-image-vs-flux-srpo-training-comparison-20-images-same-dataset/): Qwen Image base model vs FLUX SRPO training on the same 28-image dataset: 20 side-by-side results show Qwen far ahead at complex prompts and emotions. - [Qwen Image LoRA Training: Stage 1 Results and Pre-Made Configs](https://furkangozukara.com/apps/qwen-image-lora-training-stage-1-results-and-pre-made-configs/): Qwen Image LoRA results from more than 35 Stage 1 trainings with SECourses Musubi Tuner; the published pre-made configs train on GPUs with as low as 6 GB. - [Nano Banana (Gemini 2.5 Flash Image) vs Qwen Image Edit: Full Tutorial](https://furkangozukara.com/apps/nano-banana-gemini-2-5-flash-image-vs-qwen-image-edit-full-tutorial/): Nano Banana image editing tested live in 27 unique cases against the free, local Qwen Image Edit, with every prompt, image and result shown in the tutorial. - [Decoding the Timestep Shift in Diffusion Training: Qwen Image & FLUX](https://furkangozukara.com/apps/decoding-the-timestep-shift-in-diffusion-training-qwen-image-flux/): Timestep sampling shift explained for diffusion models: Qwen Image uses an exponential shift and FLUX a linear one, and 2.205 matches Qwen at 1664x928. - [Qwen Image vs FLUX Dev vs FLUX Krea: New Text-to-Image King Comparison](https://furkangozukara.com/apps/qwen-image-vs-flux-dev-vs-flux-krea-new-text-to-image-king-comparison/): Qwen Image compared with FLUX Dev and FLUX Krea over 700 generations; Qwen follows prompts far better and has a 68% bigger native resolution than FLUX Dev. - [Kohya vs OneTrainer for FLUX Dev Fine-Tuning / DreamBooth: Comparison](https://furkangozukara.com/apps/kohya-vs-onetrainer-for-flux-dev-fine-tuning-dreambooth-comparison/): Kohya and OneTrainer compared for FLUX Dev fine-tuning: OneTrainer with EMA and a copied Kohya config gave weaker results than the best Kohya config. - [RTX 5090 vs RTX 3090 Ti Speed Test: FLUX Dev, SD 3.5, SDXL & SD 1.5](https://furkangozukara.com/apps/rtx-5090-vs-rtx-3090-ti-speed-test-flux-dev-sd-3-5-sdxl-sd-1-5/): RTX 5090 benchmarked against RTX 3090 Ti on FLUX Dev, FLUX Fill, SD 3.5 Large and Medium, SDXL and SD 1.5 in SwarmUI with ComfyUI, on an AMD 9950X CPU. - [MSI RTX 5090 Trio FurMark Benchmark, Overclock & Noise Test vs 3090 Ti](https://furkangozukara.com/apps/msi-rtx-5090-trio-furmark-benchmark-overclock-noise-test-vs-3090-ti/): MSI RTX 5090 32G Gaming Trio OC review: FurMark benchmarks before and after overclocking, compared with a Gainward RTX 3090 Ti Phantom in the same PC. - [FLUX Fine-Tuning / DreamBooth Images: 2x Latent Upscale, 2048x2048](https://furkangozukara.com/apps/flux-fine-tuning-dreambooth-images-2x-latent-upscale-2048x2048/): FLUX fine-tuning / DreamBooth images made with a simple prompt and 2x latent upscale; each is 2048x2048 and training works on as low as 6 GB GPUs. - [FLUX Fine-Tuning vs LoRA: Example Dataset, Trained Models & Configs](https://furkangozukara.com/apps/flux-fine-tuning-vs-lora-example-dataset-trained-models-configs/): FLUX fine-tuning / DreamBooth vs LoRA experiment shared in full: a 28-image dataset, Kohya configs, grids and every trained checkpoint free on Hugging Face. - [FLUX De-Distilled Fine-Tuning & LoRA Tests: Bleeding and CFG Results](https://furkangozukara.com/apps/flux-de-distilled-fine-tuning-lora-tests-bleeding-and-cfg-results/): FLUX de-distilled models train best at guidance scale 3.5 instead of 1, yet these DreamBooth and LoRA experiments show they still do not fix bleeding. - [SD 3.5 Large vs FLUX Dev: Best Sampler, Scheduler & FP8 Comparison](https://furkangozukara.com/apps/sd-3-5-large-vs-flux-dev-best-sampler-scheduler-fp8-comparison/): Stable Diffusion 3.5 Large tested with 62 prompts to find the best sampler and scheduler, plus an SD 3.5 Large vs FLUX Dev comparison with raw grids shared. - [FLUX LoRA vs Fine Tuning / DreamBooth Comparison: Batch Size 1 vs 7](https://furkangozukara.com/apps/flux-lora-vs-fine-tuning-dreambooth-comparison-batch-size-1-vs-7/): FLUX LoRA and Fine Tuning / DreamBooth compared with 15 and 256 training images at batch size 1 and 7, for both realism and stylization. - [Multi-GPU FLUX Full Fine-Tuning Test on RunPod: 2x A100 80 GB Results](https://furkangozukara.com/apps/multi-gpu-flux-full-fine-tuning-test-on-runpod-2x-a100-80-gb-results/): Multi-GPU FLUX full fine-tuning / DreamBooth on 2x A100 80 GB GPUs gave only a 26% speed gain over one GPU, so 2x GPU training is not worth it yet. - [FLUX LoRA Single Block Training Research: Network Alpha & Rank Tests](https://furkangozukara.com/apps/flux-lora-single-block-training-research-network-alpha-rank-tests/): FLUX LoRA single block / layer training tested in Kohya GUI: one block needed over 10 times the usual learning rate and still gave sub-par results. - [FLUX Full Fine Tuning vs LoRA Comparison: Less Overfitting & Bleeding](https://furkangozukara.com/apps/flux-full-fine-tuning-vs-lora-comparison-less-overfitting-bleeding/): FLUX full fine tuning / DreamBooth gives way better results than LoRA training, as expected, with much less overfitting and bleeding. - [FLUX LoRA Trained on 256 Images of Myself: Mind-Blowing Results](https://furkangozukara.com/apps/flux-lora-trained-on-256-images-of-myself-mind-blowing-results/): FLUX LoRA trained with Kohya GUI on 256 phone photos of myself, including the Clip-L and T5-XXL text encoders; it learned body shape and fine facial details. - [FLUX LoRA Training With 256 Images: Full Workflow & Conclusions](https://furkangozukara.com/apps/flux-lora-training-with-256-images-full-workflow-conclusions/): FLUX LoRA training experiments with 256 images on 8x A6000 and 1x L40S GPUs: 80 epochs worked best, and doubling the learning rate for 8 GPUs overtrained. - [Kohya FLUX LoRA Training Experiments: All 104 Trainings Listed](https://furkangozukara.com/apps/kohya-flux-lora-training-experiments-all-104-trainings-listed/): FLUX LoRA training experiments run with Kohya on an 8x RTX A6000 RunPod machine, 3000 steps each, with every trained model listed and shared on Hugging Face. - [SECourses 3D Render Style FLUX LoRA Published on CivitAI](https://furkangozukara.com/apps/secourses-3d-render-style-flux-lora-published-on-civitai/): SECourses 3D Render FLUX LoRA model achieves style consistency; download it on CivitAI and get the full workflow with experiment results on Hugging Face. - [FLUX LoRA Upscale Comparison: Raw vs SwarmUI 2x vs SUPIR App](https://furkangozukara.com/apps/flux-lora-upscale-comparison-raw-vs-swarmui-2x-vs-supir-app/): FLUX LoRA output upscale comparison on an imgsli slider: the raw image vs the SwarmUI quick 2x upscale button vs our SUPIR app with face upscale on. - [Training a Public FLUX Style LoRA: Captions vs No Captions Test](https://furkangozukara.com/apps/training-a-public-flux-style-lora-captions-vs-no-captions-test/): FLUX style LoRA training in public: captions vs no captions on a 114-image 3D render dataset, 500 epochs on 4x A6000 GPUs with Kohya. - [FLUX Models Speed & VRAM Comparison: Every Version and Precision](https://furkangozukara.com/apps/flux-models-speed-vram-comparison-every-version-and-precision/): FLUX Dev and Schnell in FP16, FP8 and NF4 compared for peak VRAM usage and speed in SwarmUI on an A6000; flux1-dev-fp8 uses 19.33 GB and takes 28 seconds. - [SDXL Fine Tuning: 20 New OneTrainer Tests, Results and Better Workflow](https://furkangozukara.com/apps/sdxl-fine-tuning-20-new-onetrainer-tests-results-and-better-workflow/): SDXL fine tuning with OneTrainer on RealVis XL 4: 20 tests compare batch sizes, learning rates, xFormers, regularization images and loss weight functions. - [SDXL Training: AdamW Batch Size 30 vs Adafactor Batch Size 1 Comparison](https://furkangozukara.com/apps/sdxl-training-adamw-batch-size-30-vs-adafactor-batch-size-1-comparison/): SDXL training test on an RTX A6000: AdamW at batch size 30 vs the SECourses Adafactor config at batch size 1, which gave far better resemblance. - [DPM++ 2M SDE vs 3M SDE Sampler Test: Uniform vs Karras vs Exponential](https://furkangozukara.com/apps/dpm-2m-sde-vs-3m-sde-sampler-test-uniform-vs-karras-vs-exponential/): DPM++ 2M SDE and DPM++ 3M SDE compared with Uniform, Karras and Exponential schedules on an SDXL DreamBooth model; DPM++ 3M SDE Uniform was the favorite. - [SDXL DreamBooth Captioning Comparison: WD14 vs Kosmos-2 vs Ohwx Man](https://furkangozukara.com/apps/sdxl-dreambooth-captioning-comparison-wd14-vs-kosmos-2-vs-ohwx-man/): Image captioning in SDXL fine-tuning / DreamBooth of a single person reduces likeness with almost no benefit, in this OneTrainer test using 10.3 GB VRAM. ## News, Announcements and Support Options - [ComfyUI Latest Updates: New Models, Presets and Improvements](https://furkangozukara.com/apps/comfyui-latest-updates-new-models-presets-and-improvements/): ComfyUI update log with the latest four releases; V144 adds ten presets for Fibo 2.0, MOSS Image Edit and FastH3 Omni, for 106 ready presets in total. - [SwarmUI Latest Updates: Fibo 2.0, FastH3 Omni and New Presets](https://furkangozukara.com/apps/swarmui-latest-updates-fibo-2-0-fasth3-omni-and-new-presets/): SwarmUI update log with the four latest releases; V184 adds the Fibo 2.0, MOSS Image Edit and FastH3 Omni models and grows Amazing SwarmUI Presets to 84. - [SwarmUI Update Log: LTX 2.5 Presets, MiniMax H3 Continuation & More](https://furkangozukara.com/apps/swarmui-update-log-ltx-2-5-presets-minimax-h3-continuation-more/): SwarmUI latest updates: V172 adds multi-frame MiniMax H3 video continuation, and V170 brings 6 LTX 2.5 video presets with two-stage 1080p output and audio. - [X Algorithm Update: Deep Dive Into the Latest x-algorithm Commit](https://furkangozukara.com/apps/x-algorithm-update-deep-dive-into-the-latest-x-algorithm-commit/): X's For You recommendation algorithm got a bigger update than it looks: commit e414c17 changed 187 files with 18,263 additions across the feed stack. - [SeedVR2 & FlashVSR+ Image and Video Upscaler Pro V1 Released](https://furkangozukara.com/apps/seedvr2-flashvsr-image-and-video-upscaler-pro-v1-released/): SECourses Ultimate Video and Image Upscaler Pro 1.0 release video: SeedVR2, FlashVSR+, GAN upscalers and RIFE frame interpolation with a full queue system. - [SimplePod AI: Cheap Cloud GPU Alternative to RunPod and Vast AI](https://furkangozukara.com/apps/simplepod-ai-cheap-cloud-gpu-alternative-to-runpod-and-vast-ai/): SimplePod AI is a cheap cloud GPU service with a ready SECourses template, so all our RunPod scripts work unchanged, with permanent disks faster than RunPod. - [GitHub Account Suspension Notice: Issue Resolved, Back to Normal](https://furkangozukara.com/apps/github-account-suspension-notice-issue-resolved-back-to-normal/): SECourses GitHub account suspension notice: some scripts failed on new installs while installed apps kept working; the issue is now resolved. - [Claude AI Edits Word & Excel Files, Plus Wan Training Arrives](https://furkangozukara.com/apps/claude-ai-edits-word-excel-files-plus-wan-training-arrives/): Claude AI now edits Word documents and Excel spreadsheets; the video also shows Wan and Qwen Image training in Musubi Tuner and IndexTTS2 voice cloning. - [Massed Compute SECourses Coupon Now Works on All GPUs](https://furkangozukara.com/apps/massed-compute-secourses-coupon-now-works-on-all-gpus/): Massed Compute SECourses coupon now works on every GPU, including the newly arrived 96 GB RTX 6000 PRO, which costs 41.56% more on RunPod. - [Real Steel Became a Reality: AI Robot Boxing Tournament Explained](https://furkangozukara.com/apps/real-steel-became-a-reality-ai-robot-boxing-tournament-explained/): CMG World Robot Competition mecha fighting in 4K with English subtitles, plus how much of these humanoid boxing robots is AI and how much is human control. - [VEO 3 AI Video Generation With Native Audio: 60 User-Generated Examples](https://furkangozukara.com/apps/veo-3-ai-video-generation-with-native-audio-60-user-generated-examples/): Google VEO 3 generates short videos with native audio, dialogue and sound effects; a 4K video shows 60 wild user-generated examples from its launch. - [Deep Live Cam Windows Installer Released: Live Webcam Face Swap](https://furkangozukara.com/apps/deep-live-cam-windows-installer-released-live-webcam-face-swap/): Deep Live Cam, the viral live web cam face swap app, now has an automatic Windows installer that uses less VRAM, runs faster and supports RTX 5000 GPUs. - [Wan 2.1 Gradio App Released: T2V, I2V and V2V From 4 GB VRAM](https://furkangozukara.com/apps/wan-2-1-gradio-app-released-t2v-i2v-and-v2v-from-4-gb-vram/): Wan 2.1 Ultra Advanced Gradio app release: text to video, image to video and video to video, with the 1.3B model running in as low as 3.5 GB VRAM. - [BiRefNet HR Released: Batch Background Remover App for Windows & Cloud](https://furkangozukara.com/apps/birefnet-hr-released-batch-background-remover-app-for-windows-cloud/): BiRefNet HR brings a 2048x2048 high resolution background removal model to our upgraded batch processing Gradio app, with speed and VRAM optimizations. - [Paints-UNDO Installers Released: Undo Images Like Drawing From Scratch](https://furkangozukara.com/apps/paints-undo-installers-released-undo-images-like-drawing-from-scratch/): Paints-UNDO takes an input image and mimics how an artist could have drawn it, outputting drawing steps and a video, with a low VRAM mode that works great. - [IC-Light Image Relighting App Released: Windows, Cloud & Kaggle](https://furkangozukara.com/apps/ic-light-image-relighting-app-released-windows-cloud-kaggle/): The famous IC-Light relights images in a Gradio app more advanced than the official repo, with 1-click installers for Windows, RunPod and Massed Compute. - [LivePortrait Newest Version Released With a More Advanced Gradio App](https://furkangozukara.com/apps/liveportrait-newest-version-released-with-a-more-advanced-gradio-app/): LivePortrait newest version auto installers for Windows, RunPod, Massed Compute and a free Kaggle notebook, now with the LivePortrait Animal v1.1 model. - [Roop Unleashed Installers Released: Easy Face Swap for Windows & Cloud](https://furkangozukara.com/apps/roop-unleashed-installers-released-easy-face-swap-for-windows-cloud/): Roop Unleashed, one of the best easy face swap and deepfake apps, now has 1-click installer scripts in a separate post, with live webcam face replacement. - [Upcoming Change for New Users in 2025: Existing Members Unaffected](https://furkangozukara.com/apps/upcoming-change-for-new-users-in-2025-existing-members-unaffected/): Upcoming change for new users in 2025: the basic Bronze tier price rises for new members only, while existing monthly and yearly members keep their price. - [Crypto Payment Option: Gift Membership When You Cannot Subscribe](https://furkangozukara.com/apps/crypto-payment-option-gift-membership-when-you-cannot-subscribe/): Crypto coins can pay for a gifted membership when you cannot subscribe the normal way, for example because you are blocked or face a payment restriction. - [NVIDIA SANA Model Released: Speed and VRAM Tests, 1-Click Installers](https://furkangozukara.com/apps/nvidia-sana-model-released-speed-and-vram-tests-1-click-installers/): NVIDIA Labs developed SANA and published its weights and a Gradio demo app; with default settings an image takes 2 s on an RTX 4090 and 9.5 s on an RTX 3060. - [Start Here: All SECourses Apps, Tutorials and Scripts by Category](https://furkangozukara.com/apps/start-here-all-secourses-apps-tutorials-and-scripts-by-category/): SECourses Patreon index: an up-to-date list of every post and script with descriptions on GitHub, searchable with Ctrl+F, since Patreon has no list view. - [GitHub Account Fully Restored: Issue Resolved, Thank You for Support](https://furkangozukara.com/apps/github-account-fully-restored-issue-resolved-thank-you-for-support/): SECourses GitHub account is fully restored: an automatic security precaution had disabled it, wrongly assuming the 2FA-protected account was in danger. - [$50 for Special Help](https://furkangozukara.com/apps/50-for-special-help/): Special help for $50: one of the support options on the SECourses Patreon page, alongside the private consultations for individuals and companies. - [$250 Private Consultation for Companies](https://furkangozukara.com/apps/250-private-consultation-for-companies/): Private Consultation for Companies is the SECourses option for businesses that want direct, private help with the generative AI topics on this page. ## Guides - [AI Requirements Installer for Windows and Ubuntu: Python, CUDA and C++ in One Run](https://furkangozukara.com/guides/ai-requirements-installer-windows-ubuntu/): The AI requirements installer sets up Windows or Ubuntu for local AI apps in one run: Python 3.10-3.14, CUDA 12.9 and 13.2, cuDNN 9.20, C++ tools and FFmpeg. - [Best Local TTS Voice Cloning in 2026 With IndexTTS 2.5, OmniVoice and AuK](https://furkangozukara.com/guides/best-local-tts-voice-cloning/): Best local TTS voice cloning 2026 options compared, with IndexTTS 2.5, OmniVoice and AuK for cloning and training, RVC for conversion, and GPUs from 6 GB. - [ComfyUI 1-Click Installer With Sage Attention on Windows: Install, Launch and Update](https://furkangozukara.com/guides/comfyui-1-click-install-sage-attention-windows/): Install ComfyUI on Windows with the 1-click installer: Sage Attention, Triton, xFormers and Flash Attention on Torch 2.13 and CUDA 13, plus launch flags. - [How to Install FaceFusion on Windows, RunPod and Massed Compute (1-Click)](https://furkangozukara.com/guides/facefusion-install-windows-cloud/): FaceFusion install on Windows, RunPod and Massed Compute with 1-click installers, plus requirements, CUDA and cuDNN checks, cloud setup and common error fixes. - [How to Fix CUDA and C++ Build Errors for Python AI Apps on Windows](https://furkangozukara.com/guides/fix-cuda-cpp-build-errors-python-ai-windows/): Fix CUDA and C++ build errors in Python AI apps on Windows: set up Visual Studio, MSVC, CUDA 13.1 and 12.9.1, cuDNN, Python and PATH order the right way. - [Kohya FLUX LoRA Training Best Settings and Configs by VRAM](https://furkangozukara.com/guides/kohya-flux-lora-training-settings-vram/): Kohya FLUX LoRA training best settings by VRAM. Pick the ready config for 4 GB to 48 GB GPUs, use FP8 Scaled and block swap, and choose FLUX Dev, Krea or SRPO. - [Qwen Image LoRA Training Locally, Step by Step With Musubi Tuner](https://furkangozukara.com/guides/qwen-image-lora-training-local/): Qwen Image LoRA training locally on Windows from 6 GB VRAM. Pick a Qwen model, load a GPU preset, caption with ohwx, set epochs and find the best checkpoint. - [RunPod vs Massed Compute vs SimplePod for AI Compared](https://furkangozukara.com/guides/runpod-vs-massed-compute-vs-simplepod/): RunPod vs Massed Compute vs SimplePod for AI compared on hourly GPU prices, storage, disk speed and reliability, and how 1-click AI installers run on each. - [SeedVR2 vs FlashVSR+ Video Upscaler Compared for Speed, VRAM and Quality](https://furkangozukara.com/guides/seedvr2-vs-flashvsr-video-upscaling/): SeedVR2 vs FlashVSR video upscaler compared on the same 4K job, with speed, VRAM from 8 GB, image and video strengths and the settings that matter most. - [SwarmUI Install Guide: ComfyUI Backend, Presets and the Model Downloader](https://furkangozukara.com/guides/swarmui-install-guide-presets-model-downloader/): SwarmUI install guide for Windows: set up the ComfyUI backend, import 105 ready presets and download models in one-click bundles with the model downloader. - [How to Train a LoRA on a Low VRAM GPU (6 GB and 8 GB)](https://furkangozukara.com/guides/train-lora-on-low-vram-gpu/): Train LoRA on low VRAM GPU cards with 6 GB or 8 GB. See which models and trainers fit, how block swap and FP8 cut VRAM, and how to free VRAM before you start. - [Wan 2.2 Local Video Generation Requirements and Setup Guide](https://furkangozukara.com/guides/wan-2-2-local-video-generation-guide/): Wan 2.2 local video generation requirements explained, with VRAM per model, supported GPUs, Windows software and the fastest 1-click SwarmUI and Wan2GP setups.