--xformers CMD line option causes slightly poorer quality images on GTX 1650 (Turing), possibly others?

[cgessai]

Using the --xformers command line option on my 4GB GTX 1650 (Turing), I noticed somewhat sped-up iterations but slightly less color fidelity and quality. This is noticeable enough that I have included animated (indexed 255) color, animated desaturated grayscale .gifs, and non-xformer, and xformer images, in that order, for comparison. Especially noticeable is the apparent change of flesh tones to somewhat faded, yellowish pallor. Anyone should be able to see whether this issue affects them by rendering the same image with and without --xformers enabled and comparing the two outputs.

Non-xformer image was rendered with SD command line arguments:
--lowvram --always-batch-cond-uncond --opt-split-attention-v1 --precision full --no-half --listen

Xformer image was rendered with SD command line arguments:
--medvram --xformers

Does anyone know where I can learn more about this disparity or how to remedy it?

non-xform
with xform

[Nevoet]

I too have noticed this. Using xformers produce a slightly unsaturated images. (besides some other small details). The colors could be enhanced easily with tools like photoshop. Personally I think that the results are somewhat more realistics colorwise using xformers.

Besides that, I would like to understand the background of why is this happening. There are some cases where the image created maintains the same colors, but changes composition a little bit.

[0xdevalias]

A friend mentioned that using xformers could make things non-deterministic, and that there were a lot of references to it on the repo issues here. Wanting to understand a bit more about it, and link a bunch of potentially related issues together, I tried to find as many issues as I could that seemed to be related to xformers and the potential for it to be causing non-deterministic / unstable / inconsistent results:

The following may potentially be related (ordered by issue number):

Issues:

• Xformers optimization causes inconsistent generations with the same parameters #1999
• No token limit for prompts broken? Different results with the same prompts and seed #2005
• The same image can no longer be output under the same conditions. #2197
• On the problem that the same resolution can not be generated stably #2293
• Different results after swapping graphics card #2423
• Different images for same seed with xformers and "Highres. fix" enabled when ancestral samplers are used #2528
• Different outputs since WebUI update #2943
• xformers and euler_a variant results for same generation settings #3054
• [Bug]: The prompts do not produce the same results as before the latest updates on 10/22/22 #3476
• [Bug]: xformers always generate different results, seed is the same! #3967
• [Bug]: In img2img, setting denoising strength to 0 does not return original image #4011

Discussions:

• Different results after swapping graphics card #2418
• --xformers CMD line option causes slightly poorer quality images on GTX 1650 (Turing), possibly others? #2592
• Highres.fix composition #2705
• Exactly same prompt no more the same ? #2849
• xformers option has good and bad aspects #2958
• What has changed recently? Different and Worse Results from Same Settings #3461

I also came across this thread in the xformers repo, which while I can't guarantee is related, am wondering if it might be:

• xformers ViT-B ImageNet MAE + Deepnorm training instability facebookresearch/xformers#219

And a question I raised on a PR in the diffusers repo:

• Up to 2x speedup on GPUs using memory efficient attention huggingface/diffusers#532 (comment)

  • A friend just mentioned that apparently using xformers tends to lower the quality of generated images, and apparently even causes StableDiffusion to generate different images for the same seed/settings. I haven't looked too deeply into things to try and validate that, but I was wondering if a) that is something that is currently known/documented, and b) if that's likely to be fixed at any point, or if it's just 'the price we pay' for the benefits it adds. Would any of that be a concern with integrating things here? I did stumble upon this issue on the xformers repo, but not sure if it would be the same/similar root cause as my friend described: xformers ViT-B ImageNet MAE + Deepnorm training instability facebookresearch/xformers#219