♦ All-in-one Checkpoint fp8 is available with both fp8 and fp16 T5XXL ENCODER, choose "Full model fp16" in the downloads for fp16, and "Full model fp8" for fp8 t5xxl.
Experimental merge of SD3.5 Large and Large Turbo. I saved when I found a good balance at 8 steps, but needs further testing! So far, quality seems better than Turbo model with little speed sacrifice, although it can make some pretty incoherent images too.
Good results with CFG 1.0-3.0
Work in progress
PS. I borrowed some test prompts from SD3.5 image feed :)
Node for merging SD3.5 Large with right amount of joint blocks: https://pastebin.com/b88v5D8k
Comparisons
Description
Includes:
MODEL (FP8) • T5XXL (FP8) • CLIP L • CLIP G • VAE
FAQ
Comments (15)
looks good, even better than Flox
Could you create GGUFs? :)
work 4/8 GB VRAM ?
I’m really liking this model! The image aesthetic is awesome, way better than SDXL and Flux. The only thing is, it doesn’t stick to prompts as well as Flux does. So I might need to use Flux for the initial image, then use ControlNet or IP-Adapter. Really looking forward to when this model works with ControlNet and IP-Adapter.
The two files under 2 files have different descriptions but download the same file, they are both the 8 step one. No 16 step.
interesting success so far!
A great merge. The best so far for resource optimized SD3.5.
Can anyone advise where can I find working workflow for this model? The unet model doesn't work on FLUX work flow. And Checkpoint model doesn't work on default SD3.5 workflow - https://civitai.com/models/878934/sd35-stable-diffusion-35-workflow-upscale-simple
Please advise!!
sorry i am a noob but i am getting this error - RuntimeError: mat1 and mat2 shapes cannot be multiplied (1x1280 and 2048x2432)
what am i doing wrong?
easy and works as intended, very nice
Forge failed to recognize model type.
Can anyone advise where can I put model？thanks
v1 unet model
Does it work in Forge UI or Invoke?
Help mee.
"Token indices sequence length is longer than the specified maximum sequence length for this model (179 > 77). Running this sequence through the model will result in indexing errors"