# Photo Revival: a Codex skill turning photos into hand-drawn illustrations

> Photo Revival is an MIT-licensed Codex skill that turns everyday photos into poetic, white-paper hand-drawn illustrations through tuned prompts. It packages a specific illustration style for an agent, and it depends on an image-generation model to do the drawing.

**dacnay816y62-hub/photo-revival** — Codex skill for turning everyday photos into poetic white-paper hand-drawn illustrations.

- Repository: https://github.com/dacnay816y62-hub/photo-revival
- Stars: 581 · Forks: 57
- Language: Unknown
- License: MIT
- Published: 2026-09-18 · Updated: 2026-09-18 · Language: en
- Canonical page: https://hysenlabs.com/projects/dacnay816y62-hub-photo-revival

## What Photo Revival does

Turning a photo into a specific illustrated style usually means crafting a careful prompt for an image model and iterating until the look is right. Photo Revival packages that work as a Codex skill: it turns everyday photos into poetic, white-paper hand-drawn illustrations, a light, sketch-like style with generous white space, through prompts tuned for that aesthetic. The audience is people using Codex who want to convert photos into this particular illustrated look without engineering the prompt themselves, hobbyists, designers and content creators after a consistent hand-drawn treatment. It is a skill, an instruction package for an agent, rather than a standalone app or model, so it presupposes a Codex environment and an image-generation capability behind it. Its value is the encoded style and prompting: it captures how to get a specific poetic hand-drawn result reliably, so you invoke a look rather than reinventing the prompt each time.

## An encoded style invoked by prompt

The mechanism is a skill definition plus curated prompts that steer an image model toward one aesthetic. The repository is structured as a Codex skill, a SKILL.md with agent instructions and examples, so when you invoke it the agent applies the tuned prompting that produces the white-paper hand-drawn look, controlling composition and restraint, for instance keeping the subject small on the page with lots of white space and only a small handwritten-style annotation. You trigger it in natural language, referring to the skill and pointing it at a photo, and the skill translates that into the detailed image-generation instructions. The examples in the repository show the intended results. The design is that the hard-won prompt engineering for a consistent style lives in the skill, so the value is reproducibility of a particular look, and the actual drawing is done by whatever image model the agent drives, not by the skill itself.

## Using the skill

Photo Revival is invoked through Codex by naming the skill and describing what you want, rather than installed as a package. Once the skill is available to your agent, you point it at a photo with a request in natural language, for example:
```text
用 $photo-revival，把这张照片画成那种白纸留白的小手绘插画。
```
The skill then applies its tuned prompting to drive the image model toward the white-paper hand-drawn illustration. The repository holds the SKILL.md, an agents directory and examples, so setting it up means making the skill available to Codex and ensuring the agent has an image-generation capability to execute the prompts. The first real use is invoking the skill on a single photo and comparing the result against the repository's examples, which shows whether the style comes through in your setup, since the output quality depends on the image model behind the agent as much as on the skill's prompting.

## Where a style skill is bounded

The limitations follow from what a skill is. Photo Revival encodes one specific aesthetic, the poetic white-paper hand-drawn look, so it is purpose-built for that style and not a general photo-to-art tool; a different look means a different skill or prompt. It is a prompting package, so it does not generate images itself, it depends on the image-generation model the agent uses, and the result's quality and fidelity to the style track that model, meaning the same skill can produce different results on different backends. It presupposes a Codex environment with image generation available. And as with any style transfer, results vary with the input photo, a busy or low-quality photo may not reduce cleanly to a spare hand-drawn image. These are the natural boundaries of a single-style skill, not defects, and they mean Photo Revival delivers a look reliably to the extent the underlying model can render it.

## Photo Revival versus prompting it yourself or a general tool

The alternatives are writing the illustration prompt yourself, or using a general photo-to-art tool or filter. Doing the prompt yourself gives full control but requires the prompt engineering to nail a specific poetic hand-drawn style and to keep it consistent across photos, which is exactly the work Photo Revival has already done. A general art filter or app offers many styles but not this particular tuned aesthetic, and often less control over composition like white space and annotation. Photo Revival's difference is that it captures one carefully-tuned style as a reusable skill invoked in a sentence, so you get a consistent look with minimal effort, at the cost of being tied to that style and to your agent's image model. Prompt it yourself when you want a bespoke or different look; use a general tool for variety; and use Photo Revival when this specific hand-drawn style, applied consistently, is what you want.

## MIT license and status

Photo Revival is MIT-licensed, so it is freely reusable, which suits a skill others may adapt, and it is structured cleanly as a Codex skill with a SKILL.md, an agents directory and examples. The last push was on 2026-09-05. Adopt it when you use Codex, have an image-generation capability available to your agent, and want to turn photos into a consistent poetic white-paper hand-drawn illustration without engineering the prompt yourself, make the skill available to your agent, invoke it on a single photo, and compare against the repository's examples to confirm the style comes through on your image model. Treat it as an encoded style that delivers a specific look reliably, remembering that the drawing itself, and thus much of the quality, comes from the image model the skill drives rather than from the skill alone.

## Conclusion

Use Photo Revival if you work in Codex with an image-generation capability and want to turn photos into a consistent poetic white-paper hand-drawn illustration without engineering the prompt yourself. Do not expect it to generate images on its own or to offer other styles; it is a single-style prompting skill that drives your agent's image model, and results vary with that model and the input photo. Make the skill available to your agent, invoke it on one photo, and compare against the repository's examples to confirm the style comes through.

## FAQ

### What is Photo Revival?

Photo Revival is an MIT-licensed Codex skill that turns everyday photos into poetic, white-paper hand-drawn illustrations through tuned prompts. It packages one illustration style for an agent to apply and drives an image-generation model to do the drawing.

### How do I use it?

Invoke it through Codex by naming the skill and pointing it at a photo in natural language, for example asking $photo-revival to redraw a photo as a white-paper hand-drawn illustration. The agent needs an image-generation capability to execute the prompts.

### Does it generate the image itself?

No. It is a prompting skill that encodes a specific style; the actual drawing is done by the image-generation model your agent uses, so output quality and fidelity to the style depend on that model.

## Sources

- [dacnay816y62-hub/photo-revival on GitHub](https://github.com/dacnay816y62-hub/photo-revival)
- [Issues](https://github.com/dacnay816y62-hub/photo-revival/issues)
- [License: MIT](https://github.com/dacnay816y62-hub/photo-revival/blob/main/LICENSE)
- [README](https://github.com/dacnay816y62-hub/photo-revival/blob/main/README.md)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/dacnay816y62-hub-photo-revival