There was a bot, citripio, that used to do this for Discogs and possibly some other sites, but it hasn’t been running in a few months. Not sure if whoever runs it is planning to bring it back eventually.
Ohh please, keep Bender et al far & away from cover artwork :no_bots
Already Transhumanism is creeping me out!
Huh? My experience is the opposite. At 300 DPI I get moiré patterns (because the scanner only sees part of the dots used for printing and makes bad guesses based on that incomplete information). At 600 DPI I get the same dots as on paper. I’m curious what kind of experience other people have. I could just be wrong of course.
RIOT and Tiny PNG use lossy compression by the way. No matter how smart the lossy, it’s still lossy.
Please no, they have proven more often incorrect in my cases than correct.
Of course they use lossy compression! [quote=“outsidecontext, post:12, topic:8124”]
There was also this thread about Tools you can use to recompress JPEGs and PNGs losslessly in the old forums
[/quote]
Probably @outsidecontext didn’t mean to use the term losslessly, otherwise all these tools would be god-sents of the graphical world. The trick with tools such as, say RIOT, is to apply compression, thus reducing the filesize but without adding visual artifacts! Balancing quality and small file size used to be an art-form in itself!
Not quite correct… I don’t want to get too technical here (not my cup of tea), but here’s a definition of:
Moiré - an interference pattern created when two grids — the scanner samples and the halftone screen in the original image — have slightly different mesh sizes, or when they are overlaid at an angle.
So many scanning experts throughout the years recommend scanning at 2x the typical screen frequency (lpi)… here’s a quoted typical advice: “but that really means to scan AT LEAST AT 300 dpi. If 600 dpi is practical (meaning, if your system can survive that effort), then 600 dpi will be better.” ← So the magical solution is the higher the better? If so, we wouldn’t even be having these kind of discussions. From Wikipedia: “Some kind of moiré pattern is inevitable, but in favorable circumstances the pattern is “tight;” that is, the spatial frequency of the moiré is so high that it is not noticeable. In the graphic arts, the term moiré means an excessively visible moiré pattern.” And that’s what I mean here. My experiences lead me to say that an image scanned at 600 dpi unfiltered vs. a 300 dpi image, the moiré will be more noticeable or pronounced on the former.
The one thing that stops me from uploading cover art is the legal position.
We in the UK do not have “fair use” exemptions AFAIAA that seems to cover this in some other jurisdictions.
Perhaps others have a better perspective on this?
My impression is that if you scan at 600DPI, even if you do get moiré pattern, you can always downsample the image to whatever resolution removes the pattern/ looks best? Or is that not the case?
On that note, I haven’t really been optimising my images at all, which isn’t really my aim, but rather trying to upload the highest quality images that someone can later manipulate if they so please - eg as little information loss as possible, and then I consider my part done.
If I actually have to scan at a lower DPI to achieve that, I’ll have to change my habits I guess…
Downsampling is usually a trick employed, but doesn’t always remove the moiré viewed at 100%. If you’re a nit-picky type (and zoom in the image) you’ll still manage to see those typical dot or circle patterns, especially in dark areas.
Others say to apply a slight (gaussian) blur. This technique I never employ, firstly because it doesn’t work and secondly I love sharp images! And that’s my beef with moiré — every image in my view should be sharpened, but when you have half-tone patterns introduced, you’ll end up sharpening the artifacts, pronouncing the moiré even more.
Best solution? A good de-screening plug-in like the one I mentioned above.
I’m not saying that all scans are better at 300 dpi. I do try to scan at 600 dpi and sometimes the results are good, but most of the time the image has to be de-screened . The trade-off is blurriness, specially in type.
The main culprit is the paper source of course!
May I look at your images (images you’ve scanned)?
[quote=“AzoreanGigolo, post:28, topic:8124”]
Best solution? A good de-screening plug-in like the one I mentioned above.[/quote]It’s still not too clear to me if you’re saying that if I do this during the scanning process, it will be different to if someone else does it to my scans later. If someone else can come along and de-screen my 600dpi at any time, then I’m fine with that. My only intention is to preserve information that may otherwise be lost, not to optimise them at this stage.
Anything that’s a 7" or a CD here will have scans of mine.
Any advice on something I can improve, in relation to archival purposes as opposed to optimisation purposes, is welcome
Not during the scanning process…
“If the scanner has the Descreen or the Sharpen options, disable them. Otherwise the quality of descreening may be degraded or descreening became impossible.” (from How to Scan Printed Originals). De-screening is post-scanning. I hope that clarifies things. So my advice is for “optimization” purposes. You can read the linked manual for more tips on better scanning (archival).
BTW the scans of yours I’ve seen are of very good quality and relatively free of “screening”, good job
Concerning image size (kb) you could try RIOT to bring down the byte size. Why? Some images just take forever to load that’s all. With this program you have great control over the compression to be applied, an after & before side-by-side view (“zoomable” to pixel level) for comparison purposes. Many bytes can be shaved off without quality compromise, your eyes won’t notice the dif. I know, it’s the same argument as FLACs vs. MP3s…
Thanks for the extra info!
While it’s true that type gets a little blurrier, the descreen.net plugin keeps the parts of the image that have an actual screen pattern as sharp as possible. I think it’s something we just have to live with.
You gotta keep in mind that the amount of detail you can get out of a CD cover is very limted, anyway. Most covers only have about 150 lines per inch; if the cover measures 4.7"x4.7" there would be around 700x700 pixels on there.
At 600dpi you’re not introducing any moire or halftone patterns, just capturing what’s already there. At 300dpi you may introduce new patterns.
600dpi scans should caputure the print screen well enough to descreen most covers properly. They rarely exceed 200 lines per inch.
The right resampling algorithm sure will help keeping things smooth. I don’t see much of a difference between scanning at 300dpi and bicubic downsampling to be honest, but I guess my scanner may be downsampling or interpolating to avoid moire (and yes, the descreening option is off; I only use the descreen.net plugin).
Switching resampling to nearest neighbour will get you really bad moire patterns…