Type | |
Stats | 2,908 |
Reviews | (262) |
Published | Feb 22, 2024 |
Base Model | |
Hash | AutoV2 F5A35025A2 |
also on huggingface: https://huggingface.co/dataautogpt3/ProteusV0.4
This update enhances stylistic capabilities, similar to Midjourney's approach, rather than advancing prompt comprehension. Methods used do not infringe on any copyrighted material.
some issues have been noted "being too stylistic/creative. i am working solving those as of now and putting V0.4 into beta as to make sure people know that issues are abound regarding the model for now.
finetuned/trained on a total of 400k+ images at this point.
_______________________________________________________________________
Proteus serves as a sophisticated enhancement over OpenDalleV1.1, leveraging its core functionalities to deliver superior outcomes. Key areas of advancement include heightened responsiveness to prompts and augmented creative capacities. To achieve this, it was fine-tuned using approximately 220,000 GPTV captioned images from copyright-free stock images (with some anime included), which were then normalized. Additionally, DPO (Direct Preference Optimization) was employed through a collection of 10,000 carefully selected high-quality, AI-generated image pairs.
In pursuit of optimal performance, numerous LORA (Low-Rank Adaptation) models are trained independently before being selectively incorporated into the principal model via dynamic application methods. These techniques involve targeting particular segments within the model while avoiding interference with other areas during the learning phase. Consequently, Proteus exhibits marked improvements in portraying intricate facial characteristics and lifelike skin textures, all while sustaining commendable proficiency across various aesthetic domains, notably surrealism, anime, and cartoon-style visualizations.
Settings for Proteus
Use these settings for the best results with ProteusV0.4:
CFG Scale: Use a CFG scale of 4 to 6
Steps: 20 to 60 steps for more detail, 20 steps for faster results.
Sampler: DPM++ 2M SDE
Scheduler: Karras
Resolution: 1280x1280 or 1024x1024
please also consider using these keep words to improve your prompts: best quality, HD, ~*~aesthetic~*~