# wyrde's 4-prompt comparision Uses of [WAS](https://github.com/WASasquatch/was-node-suite-comfyui) and [omar](https://github.com/omar92/ComfyUI-QualityOfLifeSuit_Omar92) nodes. * Install the suits to avoid a mess of red boxes. ## Comparing Things Stable Diffusion can be _weird_. Sometimes it helps to examine the weirdness with pretty pictures. This worfklow runs 4 different prompts with the same settings and outputs the image results. Takes advantage of text boxes, concatenate, and cli output. ## Punctuation in Prompts ### xformers enabled | prompt A | prompt B | prompt C | prompt D |:----:|:----:|:----:|:----:| |`Is, This, Different, Than` | `Is; This; Different; Than` | `Is: This: Different: Than` | `Is. This. Different. Than` | | | | | ### xformers disabled | prompt A | prompt B | prompt C | prompt D |:----:|:----:|:----:|:----:| |`Is, This, Different, Than` | `Is; This; Different; Than` | `Is: This: Different: Than` | `Is. This. Different. Than` | | | | |
$\color{pink}{To\ disable\ xformers}$ To disable xformers: * windows portable standalone * add ` --disable-xformers` after `.\python_embeded\python.exe -s ComfyUI\main.py` in the file `run_nvidia_gpu.bat` * so it looks llke * `.\python_embeded\python.exe -s ComfyUI\main.py --disable-xformers` * github clone * add ` --disable-xformers` to the start command. * `python .\main.py --disable-xformers`
---- ## Textual Inversons (embeddings) Are the "triggers" for textual inversions required in comfyui? ### Style Princess [style princess](https://civitai.com/models/3485/princess-style) | prompt A | prompt B | prompt C | prompt D |:----:|:----:|:----:|:----:| | `girl in a sunset style-princess` | `girl in a sunset embedding:style-princess style-princess` | `girl in a sunset embedding:style-princess` | `girl in a sunset` ||||| ### GNTZ [Style GNTZ](https://civitai.com/models/22544/gntz) | prompt A | prompt B | prompt C | prompt D |:----:|:----:|:----:|:----:| | `girl in a sunset gntz` | `girl in a sunset embedding:gntz gntz` | `girl in a sunset embedding:gntz` | `girl in a sunset` | ||||| Conclusion: not needed, but they do add importance. ---- ## Lora and keywords | prompt A | prompt B | prompt C | prompt D |:----:|:----:|:----:|:----:| |`girl in a sunset `|`girl in a sunset `|`girl in a sunset holding_sign`|`girl in a sunset`| | no lora | lora 1,1 | lora 1,1 | lora 0,0 | ||||| * tested with [Anime Girls Holding Signs](https://civitai.com/models/42621/) * it's a good lora for testing since it is obvious when it is working * [compare lora workflow](compare-prompt-diff-x4-v0.8-lora.json) ---- ## resources Model * animatrix https://civitai.com/models/21916/animatrix Embedding * EasyNegative https://civitai.com/models/7808/easynegative * bad-hands-5 https://huggingface.co/yesyeahvh/bad-hands-5/tree/main Custom Nodes * [WAS Suite](https://github.com/WASasquatch/was-node-suite-comfyui) * [omar QoL suit](https://github.com/omar92/ComfyUI-QualityOfLifeSuit_Omar92)

[back] [home]