File size: 2,681 Bytes
e49d792 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 |
---
tags:
- image-to-image
- lora
- diffusers
- template:sd-lora
- ai-toolkit
- colorization
base_model: black-forest-labs/FLUX.1-Kontext-dev
instance_prompt: >-
colorize
license: unlicense
---
# COLORIZE TRUER (Variant 5)
**Experimental Version w/Color Reference Augmentation**
## Low Rank Adapter (LoRA) for Flux Kontext
## Towards Consistent + Promptless Kontext Photo Colorization
## |||| By SilverAgePoets.com ||||
<Gallery />
Our aim with this checkpoint, & other Colorize Truer/+ LoRAs, is to help restore Kontext Dev towards the far more impressive photo colorization capabilities of Kontext Pro/Max, whilst further reinforcing task-specific reliability & realism at the standards of professional manual colorization. <br>
This checkpoint was merged from several of our trained variants + an additional colorization LoRA that takes reference inputs (via typical control input channels). Alas, this color reference feature is not yet likely to function reliably/predictably. <br>
The merge was done using the sd-scripts [Flux LoRA merge script](https://github.com/kohya-ss/sd-scripts/blob/sd3/networks/flux_merge_lora.py). <br>
The constituent models were trained with [AI Toolkit by Ostris](https://github.com/ostris/ai-toolkit) over 500 grayscale/color pairings of real photos. <br>
The colorized (output) portion of our dataset consists of: <br>
Open use manual/professional colorizations of historic photos (roughly half of the data). <br>
Colorized outputs from Flux Kontext Max & Pro (the rest of the data). <br>
The source photos were sourced from public domain, and represent a broad scope of geographical, cultural, aesthetic, historical (from late 1830s thru today), & other paradigms/contexts. <br>
## Trigger words
No trigger words neccessary. <br>
With that said, with some inputs, it may be neccessary to specify `colorize` or `colorize and upscale this photo. Leave all else the same.` as the instruction prompt. <br>
One may also prompt in suggestions/corrections/directions, as far as specific choices of color. <br>
To "modernize" outputs, one may try: `Colorize and upscale the photo. Restore crisp detailing. Rich tonality. Make indistinguishable from top quality UHD 50MP+ or 12k+ pro 2025 DSLR camera colorized_02.CR2 photo. restore blurred details. Retain facial features with great fidelity source. Add fine detailing only where missing. Render real life like tones. Leave unchanged all else about photo.`, etc... <br>
## Further usage tips:
Make sure to match-up input/output dimensions, primarily to ensure fidelity of details/identity.
With starkly differing dimensions, the model might also get confused in regards to its instruction.
|