If the album art is poor quality can you scan you album art in and delete the art that is there?

Tags: #<Tag:0x00007fe320cc6a18> #<Tag:0x00007fe320cc68b0>

The covers in the musicbrainz database sure look like exact same image, just cropped differently. There’s nothing wider or narrower about the mountain tops. One is simply cut off lower.

I don’t think it needs to point to different pressings. It could be tolerances in manufacturing. Or the users cropped the images differently. Or it was cropped off because it was placed at the edge of the scannable area on the scan bed…

When cropping off the outermost edges of scans, I often have to crop off a lot, as the paper rarely is cropped perfectly in parallel with the print. Maybe it would be more accurate to pull the paper straight instead of the print, but I guess most users will much prefer to look at straightened artwork.

1 Like

i scan it in as strait as i can with out over cropping then i put it in affinity photo and delete any extra that way that extra bit matches the back ground of the player or what ever your looking at it in that makes it less noticeable

1 Like

I agree it could be either. The difference is fractional and most likely a user cropping issue. Impossible to tell without seeing the originals.

But it is funnier as you dig deeper in this example. It is very likely to be a manufacturing issue.

Look at the original vinyl at Discogs - zoomed out far enough to see the top of the mountain and blue sky above it.

Then the Mollusc scans here have a CD with text on it and the cover image has taken a bit of the mountain top off.

Then the CD at Discogs trims the top off even more (or “wider”) and has a photo on the CD.

Assuming that the images are perfect and not over trimmed by the uploaders then this could show the progression of the re-issues over the years. Each time it has been re-manufactured the image has been cropped at production.

I’ve certainly seen that occur more than once.

Impossible to know without closer inspection of the CDs. One example I’d be looking for are the SID codes that appear from the mid-1990s onwards. That will help identify which is the reissued disc here.

Currently it could be argued that both should probably have their dates removed as they are “unknown” as to which is the one released in 1989.

1 Like

If the vinyl cover was perfectly square, some cropping of the artwork is expected for the CD version, as these are usually slightly wider.

Even if we had covers of the two different versions at hand we would not be able to tell whether different cropping would affect all covers equally or if it’s due to tolerances.

Also, as we can’t see the edges of the actual cover, we can only speculate how much has been cropped off by the user or scanner. Most people place booklets on the edge of the scan bed to get it straight.

Making a good judgement here is tricky as we will have to make many assumptions.

2 Likes

The real differences here will be on the CDs and the REAR covers. Clear differences are already available between the rears visible… so I’ll be curious as to what @mmirG has on his disc and cover. Barcode on rear? Photo on disc?

That will be far more visible than the cropping on the front.

Editors removing icoverart need to be aware of the importance of minute differences between coverart that can create the basis for different Releases.

And be aware that those small differences are far more important to an encyclopedia that having high resolution scans that paper-over differences.

3 Likes

Totally agree with you. If you keep reading the thread you’ll see we are all about the little differences. It is why the rear cover is usually more important than the front cover.

This is one thing that annoys me too. Seeing people over editing images before upload. Yeah, take off scratches and damage but there is a point when over processing an image goes too far.

The images are supposed to show exactly what the Release looks like. There are plenty of other websites out there for tarted up covers for tagging needs like https://www.theaudiodb.com/.

1 Like


This image was from in hand Release. Care was taken not to change/crop the mountain top.
The apparent mountain peak is wider.

The difference between this image and the other MB frontcover cannot just be a matter of cropping by submitters as the other image is has a narrower apparent mountain top. The other submitter cannot have started off with my coverart and then un-cropped it.

2 Likes

Load all the images up at the same time. Have a look where the sailboat lines up. Look at the gap between “songs” and the top row.

Best example to look at is the LP at Discogs. That was the original photo with the full peak and sky.

I expect if all three images are loaded up in Image software they could be overlaid… which I’ll probably do in a while as I am curious. :slight_smile:

This has only ever been the one photo. Agree that manufacturing has changed how that photo has been framed.

Do you still have the physical CD? What is on your CD - photo or text? And is there are barcode on the rear?

I’m guessing that as you have the closest cropped cover then this is likely to be one of the newest versions of the release.

(Note: when I am saying “cropping” I am talking about the manufacture. I am not saying you made an error with the image.)

2 Likes

I’ve seen this happen a lot with Dark Side of the Moon or The Wall where someone finds a “pretty” image and uploads it as the cover. When clearly it is from a different version of the release. Anniversary editions often get prettified covers and someone will try and upload that to a standard version.

It is why notes are so important as to sources of images. If I see a heap of images uploaded with ZERO notes then that tells me the uploader doesn’t care. You see the pattern a lot when someone is just throwing an image at an old Release.

Edit histories are the key. It is clear when reading an edit history when a REAL editor is working as multiple images appear together, they talk about the differences, and talk about the source of the images.

I would rarely replace a scanned image. Uncredited images - more likely to go if I have scans in my hand.

-=-=-

HAHA - don’t give me stuff like this to look at :smiley:

Stick the three images up side by side. Or even better flip between then on the same screen.

Look at the text on the left. On the CD editions you can see boat sails between the word FROM. Look on the LP and you now see the red text is in a totally different place.

Compare the contour of the mountain on the Mollousc and mmirG images and it has got noticably fuzzier and less defined in mmirG’s image. (Kinda ironic for a PNG)

When looking at the top of the peak it can be see where the mmirG image cuts earlier due to the little valley on the left.

And then load up the cassette image and the text is in yet another position.

1 Like

Yeah, the quality could be a lot better. The moire issues in this scan are extreme. Scanning at 600dpi and not scaling down should give much better results.

I’ve aligned the CD covers in photoshop and the difference in crop is really subtle. Less than a mm difference.

just to say that my scan was also from my own in-hand release (obviously). i also take great care to ensure that i don’t crop out any part of the image, but i only do minimal colour correction (if any) and no “cleaning.”
could it be possible that the cover image differences are simply part of the manufacturing process? i.e. that the printed stock was physically cut in slightly different places and that they’re actually from the same release? given the very small market for this particular disc i think it unlikely that there were multiple releases, especially in just a couple of years. i can’t say definitively what the year is on my release, only that i know that it was in my possession by mid-1991 at the latest.

this issue seems to be a tangent to the original topic of this thread though.

4 Likes

Yeah, that is what everyone is saying here. :slight_smile: The good thing is you uploaded your back cover and CD images. Which lets us set the difference when compared to the copy Discogs has. You clearly have a different release to theirs. You don’t have a barcode and you have text on the disk.

That’s why I’d be fascinated to see mmirG’s disc and rear as that will then confirm those manufacturing differences. I expect mmirG’s copy to closely match the Discogs edition. (Comically you can see that the Discogs uploader has over cropped that image as they almost chopped the title off…)

As to “Off Topic” - no doubt we’ll be sliced and diced again soon. Probably at the mmirG post up there.

This is all related discussion though. It was a reason for being careful about what is replaced. It also shows that uploading as much as possible helps with clarity when later users come along with a different edition.

2 Likes

Here’s a link to images superimposed over each other:

3 Likes

Bit late, but PNG’s in as high quality as possible please!
PNG is what archivists use in my experience, and that is pretty much what we are doing.

A script can always convert them all to lossy if anyone decides that’s what MB wants, but it can’t do it the other way round. And people can automatically choose smaller jpg’s to download/tag with in Picard.

You’ve already seen that opinions are divided though so it’s up to you at the end of the day :smiley:

3 Likes

thanks aerozol i think ill keep as png at 600ppi i like quality over size as well (up to a certen point )

3 Likes

I would upload if I could find the thing.
Eventually it will appear.
Probably. :upside_down_face:

3 Likes

You checked the “no barcode” box. I bet they are identical.

Here is a jpg: https://ia800504.us.archive.org/19/items/mbid-bf79c90c-7fda-3224-a8e7-c3261b5b055b/mbid-bf79c90c-7fda-3224-a8e7-c3261b5b055b-23652402952.jpg
And here’s the png version: https://ia600504.us.archive.org/19/items/mbid-bf79c90c-7fda-3224-a8e7-c3261b5b055b/mbid-bf79c90c-7fda-3224-a8e7-c3261b5b055b-23652420227.png

The visual differences are extremely subtle, noone will be able to tell the difference.
The noise generated by the the printing process and the scanner probably far exceed the noise generated by jpg compression.

The difference in file size is huge: 8,5 MB vs 44,1 MB.

When processing the image further and working with it, the PNG will be better in theory, but in reality I really can’t notice any differene
I don’t think an offset printed cover scanned at 600dpi has enough image quality and detail that an uncompressed file could be at a real advantage.

Personally I think that we do not need to worry about jpg compression when set to the highest quality level as long as the resolution is that high, but that’s just me…

5 Likes

That is where I am going with my choice of 600dpi JPG images. When the image quality is set up at the higher end (80% in my case) then less of the data is compressed and discarded, but there is a massive saving in how the data is then encoded and stored in the file.

When the artwork for an album like Seal’s First Album is encoded ALL seven of the 2800x2800pixel images come in at less that 10MB added together. Most of that cover is just white - ideal for JPG encoding.

I believe a PNG stores data more like a BMP. So every pixel has a value. Naturally takes up more space.

Once the GUI catches up and lets us have a 1000x1000 or 2000x2000 preview we should be fine again.

Freso’s script up higher allows that to happen it adds a 1200px option witch loads fast and looks good

1 Like