@rdswift your reply is now just getting weird for the sake of it. Of course you shouldn’t create a dozen different versions. But please at least create an “average” usable version. Something cropped, rotated and somewhere within the guidelines for size. Then leave your originals as “other” for anyone who needs to go into heavier editing.
This is something I totally agree with. Not all of us can afford to own a copy of Photoshop, or understand GIMP. When nothing is there, a clean cropped scan on a home scanner is better than nothing.
Uploading these large images with colour palettes mean many people will not be able to use them as a resource.
@mmirG: “Messy edges” as an indication of cropping is fact, as you say. But saying that this should be a feature of every CAA scan…in other words that CAA scans should never be cropped…is either a policy or your preference.
@IvanDobsky: If you agree with me then you should not vote ‘no’ on rdswift’s edits, as I would argue that too is requiring arbitrary levels of correctness.
IMO the question to ask when voting is: is the MB database made better or worse by this change? I only vote “no” if I believe the answer is “worse”. Whether or not the art is usable by you in its current state, it adds documentation to the release that was not there before.
I’m the person who apparently initiated this conversation by voting against the addition of an image with colour correction info.
The primary justification for having the colour bars seems to be that it allows for the image to be corrected so that it is accurate. However, it will only ever be accurate to THAT copy of the release, at THAT time. Ink and substrate colours change as they age, and the printing process itself often creates inherent differences between items.
As such, a scanned image is not an image of a release, but of a single, specific copy of that release.
If MB info is supposed to be accurate to releases, then a scan from a specific copy doesn’t meet that criterion because it is generalised information. An assumption is made that one copy is the same as another, but this may not be true. An accurate, non-generalised image would presumably be something provided as part of the release itself, or by the publishing company, for example.
So since a scan is a generalised, copy-specific image, it can never be accurate to the release. The colour correction information is therefore unnecessary (why do you need such accuracy for something that is assumed?) and including it in cover art is adding information that is not part of the release.
If the MB database isn’t actually about accuracy but is fine with “good enough” data then the colour charts are also unnecessary, since the accuracy that they ostensibly provide is overkill (and, again, they include data that is not part of the release).
I couldn’t find out much about your color reference card, but it’s probably not too accurate as it’s screen printed and we have no way of knowing how these colors actually measure.
IT8 color targets come with a file containing reference values. These are used to calibrate the scanner, generating a color profile which gets attached to your scans. The file containing this device specific profile should then be converted to a generic working color space (like Adobe RGB) if the scan software isn’t already doing this for you. Once you’re finished with editing the file should be converted to a destination space (like sRGB for web use).
This way the image should have accurate colors when viewed on a calibrated screen and all the adjustments and technical stuff is being performed in the background.
This probably won’t get you far with your current setup, though. White and yellow on your reference card seem to be outside the scanner’s gamut (unless this is caused by your settings, but I doubt it). Also, there is some weird blurring. Edges are sharp, but large surfaces are slightly blurred, especially in dark spots. Maybe the device is set up to enhance text documents? Is this an unedited scan?
Here is a view of the clipped highlights in the Blue Oyster Cult backcover:
Any non-black pixel has one or more color channels maxed out, meaning the relationships between the colors are off here and the accuracy is compromised.
I have to admit that some of this has gone over my head. Graphical arts and digital image processing are far from my area of expertise (I’m still learning how to use Gimp). I do have an IT-8 target with a reference file, but have yet to find a way to calibrate my scanner. I’m pretty sure that the scanner is currently set with the brightness too high and the contrast a bit low. That’s about all I’ve found that I can control on the scan itself. And to answer your question, yes this is an unedited scan.
I agree with your specific post, but I voted “no” to emphasis a wider point. Every scan he is doing is in this unfinished format. I don’t get why this isn’t seen a bad practice. It really isn’t that hard to crop something to fit. Yes, the quality is good. And I thank him for doing the scans. But spending an extra two minutes creating basic versions for general use seems sensible to me. And then it makes MUCH more sense to include his original as well for those people who then want to chase for colour perfection.
As @mmirG said in the comments - these scans are like having someone’s thumb in there. It will be as good as a watermark as if I see one pop up in my collection it will make me laugh as I’ll know exactly who created it.
I’ll be watching out to see if our collections overlap. If I find any of these items downloading for my music I’ll tell you what I’ll be doing. I’ll be taking a copy of that 20MB image, and then rotating, cropping, and uploading a JPG version for other community members to make use of. Allowing users a real choice.
That was, as far as I understand, @rdswift’s entire point: he uploads these (which are good enough for at least one of the uses of cover art in MB: proving release info), and anyone is more than welcome to crop them and upload a cropped version
I’m not sure whether we need the color reference or not (the point about it being for a specific copy at a specific moment in time makes sense to me and does make it different from, say, a painting in a museum) but there’s nothing wrong as such about this upload
I’m not at all against doing this, but it is abundantly clear to me that we need to come to some community consensus of what the specs for this “usable” (for tagging purposes) version should be (e.g.: file format, file size / resolution, degree of cropping, post processing such as healing of blemishes, brightness, contrast, color balance adjustments, etc.). I don’t mind terribly doing one, but doing multiple different versions to accommodate multiple different preferences is not something that I will consider.
Perhaps another thread to develop this “community consensus” is in order. I don’t plan on starting it. I hope that I’m wrong, but I think it will be a difficult (if not impossible) task to achieve such consensus. It seems that the whole topic of cover art is extremely divisive within the community.
As a random test I downloaded and cropped one of the images. I am on a fast city based broadband at 100Mbps. Made me giggle watching it. Reminded me of the days of dialup and it appeared row by row on the screen. It took 90 seconds to appear. It then took me 60 seconds to crop and save as a JPG version.
This is what I mean. Just spend a little more effort to upload usable versions and then everyone wins. (Note I says useable, not perfect) What is currently happening will mean people with non-city broadband will fail to download it. Can Picard even handle 20MB image files? Imagine if a Picard user loads up a dozen Blue Oyster Cult albums for tagging - Picard will freeze and take ages.
If he didn’t upload it at all, they would also have to go through to process of finding a copy of the release and scanning it themselves. I’d say he’s saving them at least one step regardless of how he uploads it.
If it’s more than 0 time units, then it’s more than you can reasonably ask of them. If you don’t think it takes any time at all, you are welcome to download and process the image and upload the “usable” (by taggers? whom? different audiences have different requirements) version. It is easy (or possible at least ) to derive all the various versions you want from @rdswift’s upload, but if he had “cleaned” it, it would be very limited what would be possible to derive from it.
Apart from some cropping-and-rotating directly in the scanning software, I’m uploading unedited scans (often of library copies, with library stickers and all). Scanning the cover art and uploading it is already adding a lot of time to my editing process, and I just don’t have the time energy to spend another .5–2 (or often up to about 5) minutes on each piece of uploaded cover art, I would probably just not do it. However, like @rdswift, I’d be happy to see other community members pick up my scans and turn them into more “usable” versions.
Also, this whole discussion reminds me a bit about:
Yes it can.
Picard has no limitation on image sizes by itself, but there are limitations due to audio file formats, and of course embedding a full set of 20MB images in an MP3 of 4MB doesn’t make much sense.
You can download and save full resolution images to storage devices though.
Also note that CAA provides smaller images (250px, 500px and, since few weeks, 1200 px thumbnails), and Picard is perfectly able to use those if asked too (which is recommended, if you embed images).
Back on the topic, limitations due to formats or applications shouldn’t be taken in account, if MB and CAA are able to manage those files, that’s enough.
Does everyone agree that a checkbox to indicate “For archive purposes” is a step forward?
This would separate CA with obvious borders, colour references, existing blemishes, rips, mold, water damage, sun fading, library stickers and photographs of CA viewed through scratched jewel cases with fingers in front of the lense, from “Display for pleasure” images.
And then some way for people to choose what sort/s of images Picard etc will dish up?
If that works for the community then:
3. How about we invite all CA uploaders to add a raw scan to the CAArchive as a “For archive purposes” image as well as any “Display for pleasure” image they are generous enough to create?
That seems to be the limit of low hanging fruit.
Beyond here are monsters?
I have often wanted this option. for scans I made that were… not so fantastic. I’ve scanned a whole lot of things but I haven’t had time to fix and upload them. it would be a boon also to be able to upload some things “as is” as well as “pretty” ones for tagging. this gets a +1 from me
The files you posted don’t seem to have a color profile, so I guess your scanner is not doing any color management at all. We probably don’t need to worry about switching off color management options for calibrating.
I suggest turning the brightness down to a point that allows you to capture the black and white fields in your IT8 target without clipping (i.e. no channel should bottom out at 0 or max out at 255) and make a scan of it.
There are a couple of free software solutions for generating color profiles. Are you using Windows or Mac OS?
I’m using a windows machine for my scanning. My scanner isn’t very high quality. It’s part of an all-in-one HP Photosmart C309a series. I just posted five test scans with various brightness settings on the scanner. Are any of these of use? Thanks.
Interesting. No highlights clipping here and only two squares are clipping in the shadows. Maybe you could fix this by decreasing the contrast a little
I created a profile using coca and my IT-8 target, installed it in GIMP and tried scanning again. The result is here. Hopefully it’s a bit better. Thanks for the help.
UPDATE: It seems that Gimp wasn’t including the profile in the file when I saved it. I updated to the latest version of Gimp which seems to have saved it properly. The image file can be found here.
If other editors want to see an example of cropping of CA that leads to loss of information and possible creation of duplicate Releases then please compare
with
Somehow, and maybe not through editor action, Mick’s pupil has disappeared, as has a section of his throat.
is credited as the source at
This cropping fits inside current MB guidelines on CA, as I understand things.
There seems also to be a change in colour - this also fits insides current guidelines as I understand them.
This shrinkage of CA is what having a visible border around the outside of the CA image helps prevent. We can’t even have much confidence that the discogs image hasn’t also been cut down.
And as far as “Display for pleasure” goes, I find the warmer tones of the cropped MB CA much nicer.
:format(jpeg):mode_rgb():quality(90)/discogs-images/R-4262219-1413707145-4186.jpeg.jpg)
