Anima Turbo + CFG 1 Negative prompt
A lightweight, beginner-friendly workflow for Anima Turbo that allows negative prompting at CFG 1 using NegPip.
Turbo models are fast, but running at CFG 1 means the negative prompt box has no effect. This workflow solves that problem using NegPip, which allows you to apply negative concepts directly inside the positive prompt.
How NegPip Works
Instead of using the negative prompt box, place unwanted concepts in the positive prompt with a negative weight:
(blurry:-1)
(low quality:-2)
(text:-4)
NegPip interprets these negative weights as negative conditioning, allowing you to steer the image away from unwanted features while keeping the speed benefits of CFG 1 turbo generation.
Don't be afraid to use larger values. In my testing, weights up to ±4 can produce strong results with Anima without frying the image.
Recommended Settings
CFG: 1
Steps: 8–12
Sampler: Euler / Euler Ancestral / ER-SDE
Resolution: Up to 1536×1536 works, but ~1.5 megapixels is a good sweet spot between quality and speed in my opinion
Included Features
Anima Turbo LoRA preconfigured
NegPip integration
Impact Wildcard Encoder
Metadata saving
Simple single-pass generation pipeline
Easy LoRA and wildcard expansion
Wildcards
Create .txt wildcard files in:
ComfyUI/custom_nodes/comfyui-impact-pack/custom_wildcards
Then call them in your prompt using:
__WildcardName__
Wildcard randomization is tied to the image seed by default for reproducible generations.
Limitations
This workflow is intentionally minimal. The goal is to provide a clean, reliable starting point for Anima Turbo users who want the speed of CFG 1 generation without giving up the control that negative prompting provides.
Description
Initial release
FAQ
Comments (9)
in fact these negatives work better than CFG negatives
I didn't want to make any definitive statements since I've only been using Negpip for a few days, but yes, that's been my experience so far.
@Dioxide I've been using it for a while now, and while CFG-negatives are "suggestions" at best, this and NAG negatives have been rather "surgical"
You can sculpt the image with them, and then sculpt some more with crazy positive prompt weights; it's quite crazy, and the fact that we don't have a tiny 75 token limit, quite wonderful.
I would say negpip works better for prompt that affect local area.
If you move all negative prompt you usually use to negpip, it would give unexpected result
Bigw added support for anima in his nag fork a month ago. How does it compare? Your description of negpip makes it sound a tad less sophisticated (though probably faster since there's no separate latent).
I used NAG initially, but in addition to the slower inference, it also didn't seem to negate concepts quite as well. Granted, I didn't do any rigorous testing to compare them, but that's the feeling I got.
to sums up how both work
NAG: mix positive and negative to same embeddings
negpip: concat extra prompt you entered to the corresponding text embeddings
Works like a charm.
For anyone wondering. These same negpip node will work for SDXL and Illustrious. Checked it just in case before posting. I presume it should work for PonyXL too but check I did not.
because negpip is created for sd1.5/sdxl from the beginning