Downloads · 30 days
23
4% of all-time downloads
AlekseyCalvin/ColorizeTruerMerge4_KontexFlux_LoRA_bySAP
ColorizeTruerMerge4_KontexFlux_LoRA_bySAP is a image-to-image model from AlekseyCalvin. Use it when you need one image transformed into another. It is set up for diffusers. The card lists the license as unlicense.
Downloads · 30 days
23
4% of all-time downloads
All-time downloads
514
Public
Repo size
912 MB
Likes
0
Public
Click a slice to open those files.
.safetensors912 MB · 100%
From the Hugging Face model README
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 using the sd-scripts Flux LoRA merge script. <br> The constituent models were trained with AI Toolkit by Ostris 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 contexts <br>.
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>
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>
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.
Out of our (already quite numerous) variants of this LoRA, this merged checkpoint has exhibited the most consistent results in our tests thus far.