> This raster image must then be placed into vector software capable of performing a bitmap trace, such as Adobe Illustrator, or Inkscape.
It seems like it doesn’t actually create a vector image by itself, which is what the title had me hoping. I’d be interested to see a process that links up these two parts of the problem.
From past experience, this is the hard part. The quality has gotten better in the last 5-10 years… but until very recently, you could easily get complete garbage trying to do basic two image to vector conversion with open source tools.
I’ve been on the SVG using train for what feels like forever now, so long that I’ve tried to do the raster to vector conversion dance countless times and watched the recent progress with much satisfaction, but it’s still not perfect, two colour gradients (a common enough logo element) are frequently a problem last time I checked, which admittedly could also just be really troublesome set of examples since it’s not like I’ve got a library of images I use to regression test this stuff, I just use it.
Yes that’s the recommendation, and I’ve done it myself too… but I mentioned it as an example of what the currently available raster to vector conversion tools can’t handle.
Of particular note is when these kinds of issues crop up with overlapping elements, like having flowing ribbons or abstract line work, or other art with a colour gradient on it, which when converted to grayscale, produces the same tone or similar enough tone that the de-noising of the vectorisation decides to merge the regions as a single object. There are ways around it… but doing multiple conversions to vectorised images after performing colour space rotations, desaturations, grayscalings, and then having to stitch together a final vector in out of the correctly vectorised portions of multiple conversions… is not fun… and makes me deeply frustrated at the lack of available vectorisation software that makes full use of the colour information available.
It’s been a while since I dug down into the academic papers on the topic but when I last looked years ago, I didn’t find anything about vectorisation algorithms that didn’t basically rely on the same techniques as OCR algorithms, greyscale and contrast enhancements, before moving onto the parts of the algorithms where they began to make vectors from the raster data matrixes they had extracted.
Compared to the current state of the art in object detection in computer vision, which is admittedly something I have much more recently dug into the algorithms behind, it feels like a lot of vectorisation software is practically in the Stone Age.
It seems to be an upscaler. I guess auto-vectorizing in Illustrator works better on large images. AI upscaling has been a thing for some time now, maybe this one is trained specifically on small logos?
Was hoping this would vectorize logos but it 4x's them instead, leaving the vectorization to external tools. Maybe they should consider revising the title.
I wonder how this compares with just directly dropping the logos into adobe converter or inkscape?
Also - does anyone know of a tool that does in fact vectorize logos with AI? As in, creates sort-of semantic svg elements instead of just paths for everything - given a circle-ish thing you get a <circle> out.
I’ve had pretty good success with vectorizing raster images in Inkscape (IIRC it’s a feature called “trace bitmap”) and then doing some cleanup and adjustments. There is some manual work involved, but it really depends on the source. I was once given a small, low-resolution gif and was told that was the only copy of the logo for the government agency I was doing the work for. A few days later, I had a fully scalable SVG that became the agency’s official replacement of the logo. As an aside, Inkscape is a wonderful tool and I have used it to easily create complete icon sets (hundreds of images), which is extremely impressive for FOSS. It even has an API you can use for things like batch processing.
I haven't used Inkscape in a while but you can compare them yourself. You can use Vector Magic's web interface to rasterize an image, you won't be able do download it though.
> ESRGAN was invented to help governments see more detail with older surveillance satellites, but eventually declassified and brought into the open source world.
Is that a question or a request? If a question, the answer is no, because it's someone's website, they can do what they want. If a request, the author of the article is not in this thread and cannot provide a citation.
Yeah I'd be similarly curious how that was arrived at. ESRGAN was co-developed by a researcher at Tencent, who has been involved in a range of similar things.
The only satellite related thing I saw in my brief searching is some separate project[1] based on ESRGAN used for satellite imagery.
>This model is entirely libre (free as in freedom), and can be used for any commercial or personal purpose. My only request is that you go forth and fill the world with SVGs.
Some people release OSS out of passion. Some release it out of pain.
I had to view them in maximised full screen to see the comparison, but it looks really good. Interestingly the scaled-up version of the IBM logo has made the right edges of the B stripes be—quite understandably—flat legs of a trapezoid rather than part of the rounded B shape, but everywhere else I looked it seemed like the program scaled just as you'd want it to. Nice!
That's so cool! I've used waifu2x countless times for that exact purpose (and it did a decent job actually, but a dedicated tool should perform even better).
Tracing logos back will still give subpar results IMO – I think you're better off just using the resulting hi-res logo bitmap.
Looks great! Shameless plug: https://vectorizer.ai produces an actual vector image (I’m one of the developers of that site). Not free as in speech, but currently free as in beer.
Excellent. Wish I'd had this years ago for a bunch of projects! Tempted to revisit some of them now, knowing it can be this easy. Might be able to do a better job.
Inevitably though, the vector version will be lost to your client’s inbox.
Years from now the client will be gone and someone else from the company will explain, “I didn’t have any special design programs installed on my computer so I couldn’t open that eps thing and I didn’t know it was important.”
Save a copy and sell them bitmaps scaled as needed. Think recurring income.
Over time, I have grown to appreciate idiocy as a positive force of evolution. I now think it should be exploited whenever possible --- for the greater long term benefit of humanity.
HOLY SH*T this is absolutely amazing. I'm facing the exact situation right now with a garbage logo and no decent source. What an ingenious application, can't wait to give this a try on Monday!
It seems like it doesn’t actually create a vector image by itself, which is what the title had me hoping. I’d be interested to see a process that links up these two parts of the problem.