Comfy-Org

/

Mage-Flow

like 30

Follow

Comfy Org 10.9k

well you are using the turbo version which should have 1 cfg and 4-6 steps

For me step 1 has somewhat coherent image and then it devolves into a cloudy mess with each further step.

For txt2img turbo model, not edit.

For me step 1 has somewhat coherent image and then it devolves into a cloudy mess with each further step.

me too

For me step 1 has somewhat coherent image and then it devolves into a cloudy mess with each further step.

For txt2img turbo model, not edit.

This happened to me too, but when I tried euler sampler and linear_quadratic scheduler, the problem was solved, although the quality is still unsatisfactory.

This happened to me too, but when I tried euler sampler and linear_quadratic scheduler, the problem was solved, although the quality is still unsatisfactory.

I see some use 4 steps, but I can only use 1. With 1 step, it works, but poor quality

Try just euler/simple for the turbo, that gives the exact schedule the original code uses:

Try just euler/simple for the turbo, that gives the exact schedule the original code uses:

I could not repro this, it gives me a different image, just blobs and noise

well you are using the turbo version which should have 1 cfg and 4-6 steps.

Thanks, I forgot I am using turbo version.

1024 x 1024 seems not a good resolution.

get the much better result at 2048 x 2048

Try just euler/simple for the turbo, that gives the exact schedule the original code uses:

Euler/simple was the first thing I tried, that's where the issue is. I'm using the normal turbo model so I have omitted the TextEncodeMageFlow node and went with the default ComfyUI workflow.

Euler/linear_quadratic like Job0115 said seems to produce a proper image after 4 steps (compared to euler/simple producing clouds), I can't tell if that's the best quality the model is capable of.

I thought it might be an issue with int8 convrot so I tried with bf16 but it's still the same result.

Euler/linear_quadratic 4 steps

Euler/simple 4 steps

Euler/simple 1 step

GPU: GTX 1660 Super

I really don't know what to tell you... are you using any startup arguments?

are you using any startup arguments?

worth knowing it? --preview-method taesd --highvram --enable-dynamic-vram --disable-api-nodes --dont-print-server --disable-auto-launch

I really don't know what to tell you... are you using any startup arguments?

Other than --enable-manager, nothing else

Even with linear_quadratic it becomes messy
Euler + linear_quadratic 2 steps: low quality
Euler + linear_quadratic 3 steps: passable
Euler + linear_quadratic 4 steps: passable but more contrast
Euler + linear_quadratic 5 steps: slightly blurry and pale
Euler + linear_quadratic 6 steps: a blurry output again
Euler + linear_quadratic 7 steps: not blurry mess <----- I prefer this
Euler + linear_quadratic 8 steps: blurry outputs
Euler + linear_quadratic 9 steps: slightly blurry outputs
Euler + linear_quadratic 10 steps: slightly better outputs
Euler + linear_quadratic 11+ steps: always a mess, but remember it is Turbo intended to be used with 4 NFEs, but it should probably not behave itself like that
when edit, it adds contrast and a tint

Are all of you having the issue using GPU that does not support BF16? I know GTX 1660 Super doesn't.

I've confirmed it breaks if compute dtype is fp32, which it will be on GPUs not supporting bf16, this PR should fix the fully corrupt outputs: https://github.com/Comfy-Org/ComfyUI/pull/15081

I've confirmed it breaks if compute dtype is fp32, which it will be on GPUs not supporting bf16, this PR should fix the fully corrupt outputs: https://github.com/Comfy-Org/ComfyUI/pull/15081

my gpu doesn't support bf16. mage flow works as intended with the pr, no mess

tint and contrast change can be partially fixed with Post-Process SeedVR2 Output, the adain method (but there're better options with custom nodes to fix unwanted shift)

Just proof of concept samples for the —turbo models...

Some findings (maybe?) in very limited testing:

• euler_ancestral_cfg_pp + linear_quadratic gave good results with about 7 steps.
• er_sde + linear_quadratic as a refiner with about 2 steps seemed to clean up the image even better.
• Don't remember which ones I tried, but some other schedulers besides linear_quadratic caused the image to fall apart completely.
• In editing, er_sde + linear_quadratic with about 9 steps and CFG 1.2 stayed fairly true to the input image and corrected the replacement text.
• Perhaps a happy accident, but CLIP skip does something ...-1, -2, -3 all gave variations, while different seeds and other settings hardly did; however, less than -3 and the image deconstructed.

I've confirmed it breaks if compute dtype is fp32, which it will be on GPUs not supporting bf16, this PR should fix the fully corrupt outputs: https://github.com/Comfy-Org/ComfyUI/pull/15081

That fixed it, thanks!