TERTIUM
Experimental Turbo build consisting of a blend of my OMNIUM and ARTIUM models and 7 additional merges to achieve my first Turbo model.
The name Tertium is Latin for third a reference to this being my third basemodel, and the fact that since this is a Turbo model, it roughly takes 3 steps to get a decent output.
Technical information
The model works just as well with 1024x1024 as with 512x512, portrait aspect ratios work decently enough, but might produce slight distortions.
Requires 3 steps for good outputs, but more is always worth exploring.
Keep the CFG somewhere between 1-3, 2 is a safe bet.
The following samplers seem to work well at lower step counts with the Karras scheduler:
Euler
Heun++2
LMS
DPM++2SA
DPM++SDE
DDIM
When paired with SGM Uniform even more samplers seem to work, I like to pair that one with DPM++SDE
Description
The model works just as well with 1024x1024 as with 512x512, portrait aspect ratios work decently enough, but might produce slight distortions.
Requires 3 steps for good outputs, but more is always worth exploring.
Keep the CFG somewhere between 1-3, 2 is a safe bet.
The following samplers seem to work well at lower step counts with the Karras scheduler:
Euler
Heun++2
LMS
DPM++2SA
DPM++SDE
DDIM
When paired with SGM Uniform even more samplers seem to work, I like to pair that one with DPM++SDE
FAQ
Comments (7)
I like it
How do you merge models? I get an error If I try to merge SDXL 1.0 models wit the new turbo?
I made some grids of all currently uploaded turbo models. all but the base model and jib's need atleast 3 steps for a decent image, and anything over config 2 tends to be either very cartoony or blurry/other artifacts. also almost no difference between 1 step and 2 steps.
(step check 1-4, cfg1)https://civitai.com/posts/929205
(cfg check, 4steps)https://civitai.com/posts/929178
(same model order as pervious grids, cfg 1 steps 3) https://civitai.com/posts/929237
Very fun to work with. Recommended VAE? Or VAE baked in?
I really think this is one of the best Turbo-models right now. Too bad it's buried behind the "big names" in the listing.
Since this morning I am playing with this model and it's very good. Using a simple workflow in ComfyUI I can create very good images. Hassle free.
I wish it was not relying on negative prompts like AlbedoXL.
Seems like this model likes adding extra fingers and legs, especially if the full body isn't visible. Other than that, I'm very pleased with the results from this turbo checkpoint!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.