Considering how important Cover Art is, and how it relates to distinguishing whether two physical products are part of the same Release or different Releases (artwork differences mean different Release), there should be Style guidelines.
I propose that we get a rough consensus about some minimal Cover Art style guidelines, write a first revision of the page, and go from there. There should be links to this page from the Cover Art documentation and the How to Add Cover Art instructions.
This has been intentional in the past (we wouldn’t want to add guidelines about anything like a minimum size or scan quality). “Any difference in artwork requires a different release” (from the ticket) is not a cover art guideline, it’s a release one (which should indeed be a guideline).
I would like to keep guidelines for the CAA (so, for images/scanning as such) at a minimum. In general, I’m thinking it would be enough to have a section for cover art inside Style/Release, with stuff like:
"The cover art for a release must always exactly match the actual art for this specific release: do not add artwork to a release just because it looks “similar”. For example, digipaks are not square, and a square cover should not be uploaded to a digipak release.
Any difference in artwork requires a different release. This includes a different legal text, even if everything else is the same. An exception should be made for releases where every cover will necessarily be different (like hand-printed covers or knit covers)"
Just my 2 cents: The more detailed you force such a style guide, the less user will follow and accomplish it.
Example: Minimal cover size.
I prefer to see a tiny little matching cover then no cover at all. If I really need a bigger cover, I can search for it and compare it visually. If I have no cover, there is nothing to search and compare.
Example: Maximum cover size.
Why should a style guide restrict the cover size? If someone really has pin sharp 8000x8000px cover, why should this not be added? Everyone with little knowledge can reduce the size by itself (or the CAA does it automatically).
Silence isn’t the only way to leave style guidelines loose. We could also publish a style guideline with loose guidelines. We can have guidelines that say, some is better than none, and good is better than some. And, wrong is bad (don’t give one Release the cover art from another Release, even if it’s in the same Release Group).
From the discussion I’m seeing the outlines of a principle about what “better” is. High fidelity is good. Enough pixels to provide fidelity is good. Fewer pixels, or too many pixels, is less good than enough pixels. But contributing an image with lots of pixels in high fidelity is better than not contributing because it’s too much effort to reduce the image to fewer pixels.
It would be helpful, I think, to give guidance about whether it’s OK to use the pristine digital artwork from a digital-only Release as the cover art for the corresponding CD Release. This is essentially the “wrong is bad” case above. But maybe the pristine digital artwork matches the appearance of the CD cover, in which case maybe it’s not wrong.
Even the two paragraphs which @reosarevok quotes above would satisfy my desire for “some minimal Cover Art style guidelines”.
I’ve added a slightly extended version of my previous paragraphs to the release guidelines. I’d appreciate any suggestions on improvements to it, or any complaints if needed!
We might want to change this slightly if we decide to add STYLE-845, especially the bits about square covers But that’s easy to change.
Could we establish some guidelines or at least have a discussion about what would be the “ideal” collection of images to contribute to the Cover Art? Having this documented would be really helpful when someone tries to break up a cohesive set of scans because they think a different cover art (especially a digitally sourced one) “looks” better. Here are some ideas:
A complete cohesive set of scans: That means all cover art is from a single individual who has scanned all elements of a single issue (front cover, back cover, booklet, CD, matrix, etc.) This gives the greatest confidence in the data from an archival perspective.
High resolution: 600 dpi strikes a fine balance. It is a manageable size and any higher exceeds the limit of printed booklets. Exception: Make matrix scans at 1200 dpi because it is such a small area.
The preferred file format is a .jpg at about 85% compression. Larger files or formats don’t translate to higher quality.
No redundant versions of images (except for matrix scans of variants: any other differences in artwork would indicate a separate release).
It is preferable for large booklet and Digipak scans to be presented as a single image (or stitched together to create a single image) rather than broken into multiple images. This gives context for how the release is structured.
Trimmed vs. raw? I’m not sure about this. Maybe it would be preferable to have a trimmed “Front” image with a complete set of untrimmed (raw) images. I’m interested in getting other people’s thoughts on this.
Even this is problematic - a digipak cover that has been cropped to square, and has this flaw noted, is IMO much better than having no cover art to define the Release by.
The rest of your suggestions look good.
Physical Releases have coverart that looks different when scanned/photographed in different methods and fading/wear/soiling/marking produce very different coverart from what appears to be the same Release. IMO is often better to have this range of appearances included in the db than to have one method/faded/marked etc version presented as the accurate coverart. An exception would be when HQ images of the pristine artwork are available. But there are many Releases for which no such artwork has been found.
I agree with @mmirG that there are some occasions when it helps to have a “clean \ perfect” image of the front uploaded if the real front shots are a bit rough. I know my scanner is not always perfect, colours can be slightly out, and sometimes I can find a better image from an alternate source. (I am not a whizz at image manipulation so I upload the scans as they come from the scanner. Just some cropping\straightening)
Or the release has a perfect front, and I then add a new set of scans to cover the backs \ mediums \ booklet.
No one should be expected to scan a whole booklet. It is only really the Back and Medium we need for identification. A whole booklet is a nice bonus.
I don’t like the idea of square images when the real digibox is a different shape. I’ve seen similar square images uploaded for boxsets and it just looks odd.
I totally agree with the points 2 and 3 as this is exactly what I do. But I would never want to stop the PNG brigade as they do a great job. Just makes Picard cry when it has to download all that artwork at 22MB per image. Higher quality has its place as there was a time in the past where we thought 600x600 was enough. Look at the mess that causes at Discogs with so many unreadable rear covers because of that old rule.
6 - personally I find raw “lazy”. At least make it straight and take off most of the border. But then that is also because my usage is in a media centre. Just looks jarring to me seeing those big white areas.
5 - In a similar vein I think it is fine to do single page scans of a booklet as it is not always that easy to stitch things back together that don’t fit on a scanner. For me it depends on the time available.
An issue this brings up is the distinct lack of types needed to categorise the parts of releases. “matrix” is missing. What is the inside of a gatefold? Both vinyl and digipaks have these hard to label parts. Boxes, sleeves, slipcases. I have many types of common packaging that cannot be labelled correctly. The current types are stuck around a simple CD case with obi and sticker thrown in or single LP with inner sleeve.
I think that’s a good argument. The ideal would be a pristine, well-scanned cohesive set, but you could add an alternate “presentation” cover for the first Front scan when the cohesive set is in rough shape or poorly scanned.
I agree that no one should be expected to do this. This is just to identify the “ideal” so that editors can refer to it when making or voting on edits. However, I do think the booklet is more than a bonus. It provides a primary source for titles, lyrics, credits, etc. Matrix scans are also very useful when cross-referencing releases to Discogs.
Yeah, I wanted to see what people thought of this. When I scan, I do a quick straighten and crop leaving a little space around the artwork. It’s what I call "raw’. Then I make a second set of tightly cropped images that I use in my media player.
I use a free utility from Microsoft called Image Composite Editor. It makes pretty quick work of taking 2-6 scans and stitching them together in a single image.
What do you mean buy “inner”? We don’t yet have this type to choose.
“Tray” does have it’s own type.
One reason I think “Matrix” would be useful is to allow people tagging to leave that behind when trying to select a disc image.
I’m not saying it should delay any scans being added. Would just be nice to better categorise some of the artwork I am uploading.
Lacking anything like “inner” means all the nice bits inside every gatefold ends up being “other” or “booklet” and not really correctly categorised. This can get especially messy when I have a multi-part gatefold and having to rely on adding descriptions.
Sorry - missed the “propose” bit. I hadn’t read the rest of the replies when I got one of them notification things to confuse thing.
Yes, I like your ideas of “inner” and “Outer”, but it will also mean every front and back become “outer, front” and “outer, back”.
I can see the confusions with language already as the panel on the left on the lower image is “inside” when the package is closed. And how will you name these when they are scanned at three individual images? I call it outer left, outer middle, outer right currently.
I don’t see the need for adding “Outer” to the front and back scans. Those are well understood concepts.
Other than “Front” and “Back/Spine”, I wouldn’t scan or upload any of those other panels as individual images. Having these two images contains all the information of the Digipak/Gatefold structure and content. “Front” should be extracted as that is what most people need isolated for tagging. “Back/Spine” is useful to extract because that’s the primary reference for titles, barcodes, etc.
Have you seen some of the editors out there? There are those who like to tick “booklet” on the fronts as they argue is it the first page of a book in a jewel case.
Maybe you won’t upload single pages, but I do not always have the time and ability to reconnect the separate panels. And I know other editors are the same. There will be people who specifically want the cut up panels to make it easier to display for their own needs. There are also times when artwork is coming from a source like Discogs where there is no spine available to recreate a full outer.
Also think of a hardback book. These can’t easily be stitched together.
Your suggestion is the “ideal”, but there are a lot of people who drop outside of that who will need to be catered for. Anything that is suggested needs to cover multiple areas.
My opinion is that the stitched panels should be consider the ideal. People who want to cut them up for their own needs can easily do so. The reverse is not always possible, especially when the images are tightly cropped. Having the images cut up also makes it more difficult to see the structure of the release.
Again, cut up panels are better than none, but we’re talking about establishing an ideal, not a set of requirements for contributing scans.
But that’s easy to change.
Higher quality has its place as there was a time in the past where we thought 600x600 was enough. Look at the mess that causes at Discogs with so many unreadable rear covers because of that old rule.
