3/3/26 UPDATE
I've uploaded a custom merge of the 2603 2-step lora and my key edit of the 2602 8-step lora. This has resulted in a nice, flexible distill that can generate great outputs from 4-10 steps. It has also helped the gridmarking artifacts that were visible at 1.0 strength with the 8-step 2602 UDCAI Edit lora. Please check it out!
Original Post
Credit to alibaba-pai for their fantastic work! This is a step and cfg distill lora to be applied to Z-Image base model and its derivatives. The lora has been converted to be compatible for use in ComfyUI.
Best run at 8 steps with CFG 1. Avoid sageattention. Samplers that add noise per step such as LCM or dpmpp_SDE seem to work well.
Description
v1
FAQ
Comments (13)
they turn "base" to "turbo" right?
Not exactly. It will be 8 steps like turbo, but the outputs will be very different. Base will have better flexibility for styles and art beyond photoreal.
It's basically ZBase at ZTurbo speed. More output diversity.
Use this lora on standart ZIT - greatly improves the quality.
Also try strenght 2.0
Should the weight still be 0.7-0.8 as the original?
Yes, in my testing I would say weighting around 0.6-0.8 is probably best. 1 certainly works but it has a tendency to appear too overexposed or burned.
@UDCAI Also, I did try LCM which seems to work well but dpmpp_sde seems overexposed or plastic. This was at lora weight 0.8. Which scheduler were you using with sde?
@EsheAI I mostly use bong_tangent
0.6 seems to work best with on-site Euler sampler and no other LoRAs for me
@GlossyMannequin I didn't realize the online gen was so limited. I did some quick tests with lora at 0.5 strength, 1.5cfg, at 12 steps euler simple. This seems to be the most consistent.
@UDCAI I tested several step counts and CFGs on 0.5 strength (same seeds), and 12 steps at 1.5 seems actually on par with 10 steps on 1, which is 20% cheaper in terms of buzz
Details
Files
Z-Image-Fun-Lora-Distill-8-Steps_ComfyUI.safetensors
Mirrors
Z-Image-Fun-Lora-Distill-8-Steps_ComfyUI.safetensors
Z-Image-Fun-Lora-Distill-8-Steps_ComfyUI_v1.safetensors
Z-Image-Fun-Lora-Distill-8-Steps_ComfyUI.safetensors
Z-Image-Fun-Lora-Distill-8-Steps_ComfyUI_v1.safetensors
Z-Image-Fun-Lora-Distill-8-Steps_ComfyUI.safetensors
Z-Image-Fun-Lora-Distill-8-Steps_ComfyUI.safetensors
Distill-8-Step-Lora-b.safetensors
Z-Image-Fun-Lora-Distill-8-Steps_ComfyUI.safetensors
Z-Image-Fun-Lora-Distill-8-Steps.safetensors
Z-Image-Fun-Lora-Distill-8-Steps_ComfyUI.safetensors
Z-Image-Fun-Lora-Distill-8-Steps_ComfyUI_v1.safetensors