Image
interior-designer-ai
Upload a sample room photo plus get a design back in seconds.
★ 215 starsMIT
View guide →InfiniteYou (InfU) — flexible photo recrafting on FLUX that preserves your identity (ICCV 2025 Highlight)
InfiniteYou is an open-source image project by bytedance (bytedance/InfiniteYou on GitHub). It has 2,685 GitHub stars and 286 forks, is released under the Apache-2.0 license, and was last updated on GitHub on 22 August 2025.
Prepare one clear identity photo and the prompts for the new photos you want, and check your GPU VRAM against the memory options.
Clone the repo; model weights download from Hugging Face (ByteDance/InfiniteYou) with FLUX.1-dev as the base model.
pip install -r requirements.txt, and choose memory settings: full bf16 needs about 43 GB peak VRAM per the README, or combine --cpu_offload and --quantize_8bit to get to about 16 GB.
Run test.py with your ID image and a prompt, then compare the aes_stage2 variant (better aesthetics) with sim_stage1 (higher identity similarity).
Identity features are injected into the FLUX DiT via residual connections, preserving identity while keeping generation flexibility.
aes_stage2 (after SFT; better text alignment and aesthetics, the default) and sim_stage1 (higher identity similarity), switchable with --model_version.
Pretraining plus supervised fine-tuning with synthetic single-person-multiple-sample (SPMS) data, addressing face copy-pasting.
Works with FLUX variants like FLUX.1-schnell, ControlNets, LoRAs (optional Realism and Anti-blur LoRAs are provided), IP-Adapter and OminiControl.
--cpu_offload and --quantize_8bit flags reduce peak VRAM substantially with similar quality, per the README's measurements.
Local inference script, local Gradio demo (python app.py), an online Hugging Face demo, and official ComfyUI nodes.
Clone InfiniteYou and enter the directory.
git clone https://github.com/bytedance/InfiniteYou.git
cd InfiniteYouInstall the requirements as the README instructs.
pip install -r requirements.txtGenerate from an identity photo and prompt with the README's quick-inference command.
python test.py --id_image ./assets/examples/man.jpg --prompt "A man, portrait, cinematic" --out_results_dir ./resultsByteDance Intelligent Creation researchers (Jiang, Yan, Jia, Liu, Kang, Lu) maintain it; the paper appeared at ICCV 2025 as a Highlight.
The code is Apache-2.0, but the model is CC-BY-NC 4.0 — academic research only, non-commercial. FLUX.1-dev and InsightFace face models carry their own licenses too.
No — it runs locally, but it is GPU-heavy: about 43 GB VRAM at full precision, or about 16 GB with offload + 8-bit quantization per the README.
The paper's comparison reports stronger identity similarity, text alignment and aesthetics than those baselines, and less face copy-pasting.
Clone, pip install -r requirements.txt, then run python test.py with --id_image and --prompt, or try the Hugging Face demo first.
Upload a sample room photo plus get a design back in seconds.
Open-source alternative to the popular AI photo editor — all local
Cloud-native AI photo editor — cutout, batch edits, object removal, HD upscale
A self-hosted open source photo management service
Create LoRA training datasets in the browser — Flux 2, Qwen, Nano Banana
Open-source multi-brand RAW photo editor with AI enhancement and smart masks
The most-starred other Image projects in the LibreAI catalogue — the realistic shortlist if InfiniteYou is not the right fit.
Stable Diffusion web UI
The node-graph powerhouse for precise, professional AI image and video generation.
High performance self-hosted photo and video management solution
Focus on prompting and generating
🆙 Upscayl - #1 Free and Open Source AI Image Upscaler for Linux, MacOS and Windows
Free and open — grab the source, star the repo, and start creating.