lucataco / proteus-v0.1

ProteusV0.1 uses OpenDalleV1.1 as a base and further refines prompt adherence and stylistic capabilities to a measurable degree

  • Public
  • 6.6K runs
  • GitHub
  • License



Run time and cost

This model runs on Nvidia A40 (Large) GPU hardware. Predictions typically complete within 22 seconds. The predict time for this model varies significantly based on the inputs.


Cog implementation of dataautogpt3/Proteus-v0.1


ProteusV0.1 is currently my best model to date.

ProteusV0.1 uses OpenDalleV1.1 as a substantial base to work from in terms of an untuned base model. It further refines said prompt adherence and stylistic capabilities to a measurable degree. tuned on 10k unique examples of high-quality captioned images generated from dalle3 and then further refined using 220k GPTV captioned non-copyright stock images with some anime through in here and there.

ProteusV0.1 has a noticeable improvement in facial features and skin detailing. while still keeping similar or slightly improved levels of surrealism, anime, and cartoonish styles.

Settings for ProteusV0.1

Use these settings for the best results with ProteusV0.1:

CFG Scale: Use a CFG scale of 8 to 7

Steps: 20 to 35 steps for more detail, 20 steps for faster results.

Sampler: DPM 2M++

Scheduler: Karras

please also consider using these keep words to improve your prompts: best quality, HD, aesthetic,