Hacker News Reader

Comment by prawn | Hacker News Reader

Hopefully related: If I'm a photographer wanting to improve resolution of my content for printing, what's my current best bet for upscaling?

Is it realistic to make use of this on the command line, feeding it my own images? Or has someone wrapped it in an app or online service?

View on HN

You're probably better of using Real-ESRGAN: https://github.com/xinntao/Real-ESRGAN. It's pretty solid and fast, even has portable executables you can just use as is. The upscaler that comes with stable diffusion might work for you, but I suspect it'll probably do a better job at upscaling stable diffusion output rather than a natural images (might be wrong though).

Not sure about your last claim. The example given in the blog post looks very very close to a natural image.

I tried this on one of my home images. I have a nice canon pro 100 printer that can print 13”x19” pictures, and my camera is a 20 megapixel Panasonic GH-5. The printer can print much higher resolution than my camera. So I did take one of my photos and process it with Real-ESRGAN to double the resolution (in each direction, so 4x pixels). The photo is a red barn with redwood trees behind it. It did well increasing the resolution of the barn. It made it look more crisp and bright. But there is an area with some trees in shadow behind the barn, and it lost detail there.

Anyway I think it would be fun to play with, just depends on the content of the image and the artists preferences. I still haven’t printed a full page of the upscaled photo but I do want to try that and see how it looks in comparison!

As a counter-recommendation, Topaz’s much-advertised Gigapixel AI is rarely useful. Their Denoise and Sharpen apps are good though.

I dunno - I've found it useful on a bunch of images[1] but I tend to try Pixelmator Pro first because that's a simple key combination to enlarge an image and 90% of the time it's Good Enough for my purposes.

[1] The new Photo AI, on the other hand, is slow, clunky, and not infrequently glitches out wildly. But on the plus side it does combine sharpening and denoising into one workflow.

I was super unimpressed with the 1.0 release of Photo AI; in particular, the sharpening was a LOT slower than standalone. But that's fixed now, and unless Topaz starts backporting the improved models to the standalone tools -- so far, they have not -- Photo AI will get you better results.

I spent a lot of time last month using Gigapixel (actually the improved version in their new Photo AI product) last month on dozens of images for my dad's memoir. There were a couple failures where the input image was just so blurry or low-res that it couldn't be saved, but Topaz significantly improved image quality while upscaling in 90+% of cases.

If you have a Nvidia GPU, I've been using Upscayl locally and for free with decent results: https://github.com/upscayl/upscayl

Note that on some image types it tends to make things look digitally painted rather than detailed. I recommend you try a few different tools and see what works best for the type of photography you do.

This thread has a useful app. https://news.ycombinator.com/item?id=32628761

Photoshop and Lightroom have had AI upscaling for awhile.

Ah - had forgotten. I'll try them first. Thanks.

So does Pixelmator. You can try the free trial which comes with this feature.

Pixelmator's Super ML Resolution does a great job with upscaling images, can highly recommend it.

Yeah, I use it quite a lot on stuff generated by Midjourney and the results are always great.

Reply on news.ycombinator.com