DDColor is a free image-colorization model for restoring color to historical black-and-white photos and recoloring anime-game landscapes. Its approach combines multi-scale visual features with learnable color tokens to produce automatic colorization. The repository provides paper, ModelScope, artistic, and tiny pretrained variants; its maker recommends ModelScope for images outside ImageNet and describes tiny as the lightest option. Local inference scripts and a runnable Gradio demo are documented, alongside inference guidance for Hugging Face Hub and ModelScope. The repository also links to Replicate demos and an API. Users can train with ImageNet or a custom dataset, and the project documents ONNX export with an example using the tiny model. Requirements are Python 3.7 or later and PyTorch 1.7 or later. The repository includes an Apache License, Version 2.0. One limitation noted by the maker is that colorfulness loss can create unpleasant blocks of color, including red artifacts.
Who it is for
DDColor suits developers or image-workflow users looking for a self-hosted colorization model, with local inference, training, and export options. It requires Python 3.7 or later and PyTorch 1.7 or later.
What is good
- Free model with four pretrained variants
- Includes local inference scripts and a Gradio demo
- Training supports ImageNet or custom datasets
- ONNX export is documented
What to know first
- Colorfulness loss can create unpleasant color blocks
- Requires Python 3.7 or later and PyTorch 1.7 or later
Verdict
DDColor offers multiple pretrained options and documented local inference, training, and export paths. The possibility of color artifacts is worth considering for image-colorization work.
Get started with DDColor
- Open the DDColor repository at https://github.com/piddnad/DDColor.
- Choose a pretrained variant; ModelScope is the maker’s default recommendation for images outside ImageNet.
- Prepare an environment with Python 3.7 or later and PyTorch 1.7 or later.
- Follow the repository’s local inference instructions or run its Gradio demo.
- For hosted inference, use the documented Hugging Face Hub or ModelScope instructions, or explore the linked Replicate demos and API.
Questions about DDColor
How much does DDColor cost?
The listed plan is free at no cost, with no paid plans or product usage limits stated.
Which platforms are listed?
The listed platforms are API, Linux, and self-hosted.
Which pretrained models are available?
The repository provides paper, ModelScope, artistic, and tiny variants. The maker recommends ModelScope by default for images outside ImageNet and describes tiny as the most lightweight.
Can I train or export the model?
Training is supported with ImageNet or a custom dataset. The repository documents ONNX export and includes an example using the tiny pretrained model.
What license does DDColor use?
The repository includes the Apache License, Version 2.0.
What can cause unwanted color results?
The maker cautions that colorfulness loss can produce unpleasant color blocks, including red artifacts.
DDColor plans and pricing
All plansCompared on AI photo colorization software
- Free plan
- Yes
- Batch processing
- Yes
- API access
- Yes
- Commercial license
- Yes
