Nodes Browser

ComfyDeploy: How ComfyUI-ppm works in ComfyUI?

What is ComfyUI-ppm?

Fixed AttentionCouple, NegPip(negative weights in prompts) for SDXL and FLUX, more CFG++ and SMEA DY samplers, etc.

How to install it in ComfyDeploy?

Head over to the machine page

  1. Click on the "Create a new machine" button
  2. Select the Edit build steps
  3. Add a new step -> Custom Node
  4. Search for ComfyUI-ppm and select it
  5. Close the build step dialig and then click on the "Save" button to rebuild the machine

ComfyUI-ppm

Just a bunch of random nodes modified/fixed/created by me or others. If any node starts throwing errors after an update - try to delete and re-add the node.

You can drag-and-drop workflow images from examples/ into your ComfyUI. I'll probably add some more examples in future (but I'm kinda lazy, kek).

Nodes

CLIPNegPip

Supports SD1.5, SDXL and FLUX!

Modified implementation of NegPiP by laksjdjf and hako-mikan. It uses ModelPatcher instead of monkey-patching, which should increase compatibility with other nodes.

CLIPNegPip node allows you to use negative weights in prompts. Connect the node before other model/clip patches.

Read more about NegPiP in the original repo. I recommend to keep all dots/commas inside weight braces (i.e. (worst quality,:-1.3) (sketch:-1.1,) instead of (worst quality:-1.3), (sketch:-1.1),).

[!NOTE] CLIPNegPip is compatible with:

CLIPNegPip is incompatible with:

Example workflow (FLUX)

workflow_negpip_flux

AttentionCouplePPM

Modified implementation of AttentionCouple by laksjdjf and Haoming02. I made AttentionCouplePPM node compatible with CLIPNegPiP node and with default PatchModelAddDownscale (Kohya Deep Shrink) node.

Inputs for new regions are managed automatically: when you attach cond/mask of a region to the node, a new cond_ / mask_ input appears. Last cond_ / mask_ inputs are always optional.

Use multiple LatentToMaskBB nodes to set bounding box masks for AttentionCouplePPM. The parameters are relative to your initial latent: x=0.5, y=0.0, w=0.5, h=1.0 will produce a mask covering right half of your image.

DynSamplerSelect

Modified samplers from Euler-Smea-Dyn-Sampler by Koishi-Star.

Contains some new samplers: euler_ancestral_dy, dpmpp_2m_dy and dpmpp_3m_dy.

Tweaking s_dy_pow may reduce blur artifacts (optimal value is 2 for euler_* samplers and -1 for dpmpp_* samplers, use -1 to disable this feature).

CFG++SamplerSelect

Samplers adapted to CFG++: Manifold-constrained Classifier Free Guidance for Diffusion Models by Chung et al.. Also contains converted samplers from Euler-Smea-Dyn.

Should greatly reduce overexposure effect. Use together with SamplerCustom node. Don't forget to set CFG scale to 1.0-2.0 and PAG/SEG scale (if used) to 0.5-1.0.

Guidance Limiter

Implementation of Applying Guidance in a Limited Interval Improves Sample and Distribution Quality in Diffusion Models by Kynkäänniemi et al. (also contains RescaleCFG functionality)

Guidance Limiter is also available as a CFGLimiterGuider guider node for SamplerCustomAdvanced.

Empty Latent Image (Aspect Ratio)

Empty Latent Image (Aspect Ratio) node generates empty latent with specified aspect ratio and with respect to target resolution.

CLIPTextEncodeBREAK

A small lightweight wrapper over ConditioningConcat node, CLIPTextEncodeBREAK node can split prompts by BREAK keyword into chunks and produce a single concatenated conditioning.

CLIPTokenCounter

Counts tokens in your prompt and returns them as a string (currently limited to clip_l). You can also print token count + individual tokens by enabling debug_print.

Hooks/Hijacks

Schedulers

Adds AlignYourSteps scheduler modified by Extraltodeus to the default list of schedulers by replacing comfy.samplers.calculate_sigmas function. ays is the default AYS scheduler for SDXL and ays+ is just ays with force_sigma_min=True.

Also adds GITS scheduler and AYS_30 scheduler (based on AYS_32 by Koitenshin)

Advanced CLIP Text Encode

Hijacks advanced_encode_from_tokens method from Advanced CLIP Text Encode extension (if installed), making all weight interpretations compatible with NegPip.