UnCanny (Photorealism Chroma)

UPDATE V1.3: Aims for a middle ground between v1 and v1.2. Hopefully fixes the biggest issues of both versions. Pick bf16 or fp8 on the right ----->. GGUFs on HuggingFace.

Chroma is a fantastic and highly versatile model. This finetune aims to improve reliability in realistic/photo-based styles while preserving Chroma’s broad concept knowledge. The v1.3 flash version has the rank-256 lora (from here) baked in. Prompting: Simple prompts describing what you want to see in natural sentences works well. Chroma prompts work well. Examples of captioning style used in training: amateur-guitar, night-sky, close-up face, tiger. Negative prompts do not work at CFG one. With CFG above one, negative prompts can be very important (for good or bad). Example settings (not necessarily optimal):

• Steps (base): ~30-40 (depends on other settings; CFG, sampler, etc.)

• Steps (flash lora): 15-17 works well with rank-128/256. Depends on lora rank.

• CFG (base): ~3.5 (depends on other settings; steps, sampler, etc.)

• CFG (flash lora): 1 works well with rank-128/256. Depends on lora rank.

• Sampler: res_2m and dpmpp_sde work well and are used for example images.

• Scheduler: I like bong_tangent | beta is also good.

Support:
Have too much money? Want to support further training? https://ko-fi.com/dawncreates

Training Details
The model was trained locally, using Chroma-HD as the base. Each epoch included images at 3–5 different resolutions, though only a subset of the dataset was used per epoch. Except for the extra resolutions, OneTrainer's default config for 24gb Chroma finetuning was used. The dataset consists almost exclusively of SFW-images of people and landscapes, so to retain Chroma-HD's original conceptual understanding, several layers were merged back at various ratios. All the juice, compositions, subjects, and concepts come from Chroma itself, my model just nudges it towards realism. Honestly, this version is more of a showcase of how good Chroma is than a great finetune in itself. I do think it shows how much potential Chroma has for finetuning though - so get to work on Chroma finetuners - it has so much potential!

All images were captioned using JoyCaption: https://github.com/fpgaminer/joycaption

The model was trained using OneTrainer: https://github.com/Nerogar/OneTrainer