stable-diffusion.cpp
stable-diffusion.cpp copied to clipboard
leejet
Diffusion model(SD,Flux,Wan,Qwen Image,Z-Image,...) inference in pure C/C++
It would be great if support for the SDXS-512 model could be added: https://github.com/IDKiro/sdxs Especially for quickly generating images on the CPU, this is a major new development, so it...
When I tried out the released binaries, I was surprised they do not have flash attention enabled. I mean these: [sd-master-ce1bcc7-bin-win-avx-x64.zip](https://github.com/leejet/stable-diffusion.cpp/releases/download/master-ce1bcc7/sd-master-ce1bcc7-bin-win-avx-x64.zip) [sd-master-ce1bcc7-bin-win-avx2-x64.zip](https://github.com/leejet/stable-diffusion.cpp/releases/download/master-ce1bcc7/sd-master-ce1bcc7-bin-win-avx2-x64.zip) [sd-master-ce1bcc7-bin-win-avx512-x64.zip](https://github.com/leejet/stable-diffusion.cpp/releases/download/master-ce1bcc7/sd-master-ce1bcc7-bin-win-avx512-x64.zip) The readme mentions this about flash attention:...
Enabling Flash Attention completely breaks prompt following, there has to be a bug in the flash attention implementation. I reproduced this with the simplest possible setup, two identical locally compiled...
Hi! There is a problem with the mac os build. When running ./sd binary, it outputs an error ```log dyld[37566]: Library not loaded: @rpath/libstable-diffusion.dylib Referenced from: sd-master--bin-Darwin-macOS-14.4.1-arm64/sd Reason: tried: '/Users/runner/work/stable-diffusion.cpp/stable-diffusion.cpp/build/bin/libstable-diffusion.dylib'...
Seems to be an issue with group_norm on metal, haven't tried with other backends. ``` ./bin/sd -m models/gsdf/Counterfeit-V2.5/Counterfeit-V2.5_pruned.safetensors -p "a cat" --steps 2 -H 640 -W 640 ggml_metal_init: allocating ggml_metal_init:...
I tried to preprocess_canny a bitmap with a size of 1024x1024. Due to the hardcoding of the memsize, ``` uint8_t* preprocess_canny(uint8_t* img, int width, int height, float high_threshold, float low_threshold,...
For instance, some effects require combining multiple different control networks to achieve them, which is easy to implement in ComfyUI. However, this is not possible with the current Stable Diffusion...
Recently i discovered Dllama.dll ( https://github.com/tinyBigGAMES/Dllama ), lib that allows to chat with llm ggufs using only one simple function call. For example, i call it from my AutoIt script...
It would be great to add input parameters to the current SD cli to specify an input and mask file to run the inpainting. For example: `./bin/sd -m ../models/sd-v1-4.ckpt -p...
This PR will probably take me a very long time since I'll have to make many modifications to the code. Due to changes in the way of building the computation...
