I've talked a bit about why I restore images before, and, in a recent issue, covered the research process I go through to find the images.

But perhaps I haven't really shown how big of a role researching images alone plays in improving Wikipedia. Even if you feel image restoration is beyond you, it's often not particularly hard to improve Wikipedia in a substantial way by research alone. So let's not just look at images before and after restoration — let's look at what was in the article before the research.

Nettie Stevens was one of the two independent discoverers of the Y-chromosome, and arguably had much deeper insight into its role. Her before-research image is fairly typical of the sort of very mediocre image that can readily be replaced. As she's American, I tried the Digital Public Library of America (DPLA), at https://dp.la, which found two images of her: one a much better copy of the image in use, and the other of which obscured her face with a microscope. Then I restored what I found:

• 
  Original image in article
• 
  Image I found
• 
  Image after restoration

The biggest jump in quality, in my opinion, is between the old image in the article and the better copy I found through DPLA. The original looks like a photocopy. The research is more important than the restoration here, and Wikipedians are generally good at research.

The other, less insightful, independent discover of the Y-chromosome was Edmund Beecher Wilson. When I was cleaning up the Nettie Stevens images, I thought perhaps I should add his image to XY sex-determination system as well as hers, went to his article, and realised something had to be done. As he was also American, I used DPLA again, and found good enough results I didn't need to go further:

• 
  Original image in article
• 
  Image I found
• 
  Image after restoration

There's probably a bigger boost from the restoration here, but at the same time, the original is so appalling that one would be hard-pressed to say that the restoration is the bulk of the improvement.

Also, this image brings up issues with the specifics of restoration. Now, archives go offline all the time, so I reiterate my position that you need to upload the unaltered original. But you don't need to use it in articles. The biggest improvements at thumbnail level were probably the crop, the levels adjustment, and fixing the damage to the edges that show up as areas of white. The first anyone could do, the second requires a little care (it's very easy to go too far), and the third is much more difficult; but that does mean with a little care, anyone could have done two of the most substantial improvements.

The original lead image for Emily Davison isn't too bad at thumbnail size, but it's pretty appalling zoomed in: it looks like an upscale of a much smaller image. I used either Bing or Google to find this one; I don't remember which, as I often try both and see which one I get better results with. As often happens with researching an image, it conveniently turned an unknown photographer into a known photographer: in this case, Andrew William Dron. Who was a research project in himself to get a birth and death year for, but that's maybe beyond the scope of this article.

• 
  Original image in article
• 
  Image I found
• 
  Image after restoration

I think this one shows the benefits of research the best, since, at thumbnail, outside of the crop, there's little visual distinction. It's pretty obvious that there's a bunch of white spots that got removed once you zoom in a little more, but thumbnails are rather forgiving:

• 
  Image I found
• 
  Image after restoration

...And that's to your advantage. It means that a host of small flaws will be forgiven at the article level. Now, Wikipedia articles are widely used to find images of people nowadays. So there is a good reason to fix them, if you can. But even if you can't, you're still doing a very valuable job of putting substantially improved images out there that can be readily reused. Don't let the perfect be the enemy of the good.