Image capture renders text wrong size for some users
Reported version
3.0
Priority
P0 - Critical
Type
Functional
Frequency
Once
Severity
S3 - Major
Reproducibility
Always
Status
closed
Regression
No
Workaround
No
Project
When using the "camera" screen-capture tool in MS 3.0, measure numbers, should you have them enabled, are rendered to the image at egregious size. See enclosed.
Attachment | Size |
---|---|
MS3MnoFail.png | 46.39 KB |
Fix version
3.0.1
Comments
I just see 2 red dots on black backgroud?
Ah! It is Chrome rendering it badly, how about this:
I can't reproduce though, here's it how it renders measure numbers for me:
In reply to I just see 2 red dots on… by Jojo-Schmitz
I wish MS would not make transparent images as it does, or at least optionable. It just hit you.
So please share a score showing the issue.
What OS?
In reply to I can't reproduce thoigh,… by Jojo-Schmitz
Maybe it's because I'm in continuous scroll mode?
In reply to Maybe it's because I'm in… by [DELETED] 1831606
No, it happens in page mode, too.
You can disable transparency in Preferences/Export
In reply to No, it happens in page mode,… by [DELETED] 1831606
I am using "measure numbers every 1" (as I continually advise almost everyone to do) ... maybe that.
Nope:
In reply to You can diesbla transparency… by Jojo-Schmitz
... had forgotten doing that in MS2 years ago.... thanks.
The score was created in MS2; maybe that's the issue.
No, that is true for me sample score too
So please share a score showing the issue.
What OS?
In reply to (No subject) by Jojo-Schmitz
Mac Mojave.
Here you go. That's it, import from MS2.
In reply to (No subject) by Jojo-Schmitz
What info needed? Does MNFO in MS3 not do it for you?
Not sure yet, but I've marked the issue active, as for now all request info have been given
I still can't reproduce:
Windows 7, Beta as well as latest development build
In reply to I still can't reproduce: … by Jojo-Schmitz
Try Mac. Happens every single time for me.
Send enough money and I'll buy one ;-)
Marc could reproduce
I wished I could buy Windows, I'm using Linux ;-).
Maybe seems to be a Mac issue, isn't reproducible with Linux too:
Happens every time for me too, not just measure numbers but also staff text, I'll guess any text. And I'm on Windows.
My theory is it relates to screen scaling for high DPI displays. I also have palettes that are displaying smaller than in 2.x, not everyone sees that. How about you?
In reply to Happens every time for me… by Marc Sabatella
My palettes look to be the right size (although I'm ambivalent about the new font).
Shouldn't be a new font in general, fwiw, although depending on your system configuration, you might see something different. previously we just used the default system font & size, now we give controls in Edit / Preferences, and depending on whether your system default match what we are now setting, it may or may not match what you were seeing before.
Anyhow, do you know - or can you calculate - your monitor's DPI? Is it a "retina" display? Also, can you verify the problem is all text, all scores?
I do have a retina display, 227 px/in. It does happen for staff text, too. What other types of text would you like to hear about? I would think one repeatable, brand-new failing score would be sufficient (I have hundreds of MS2 scores).
If it helps any, I have a score...
...made from scratch using the String Quartet template that shows this.
...made by using Save as... in a section of empty measures from a 2.3.2 score that shows this. (I don't remember if I used a template for this one).
...imported from 2.3.2 using a symphonic template and accepted reset of defaults that shows it.
I don't normally use the capture tool, so I haven't noticed this before. I tested this on 3 scores from different sources all with the same results.
Thanks, no more scores needed :-). I am 99% sure it's not score-dependent but system-configuration-dependent. That is, those of us who see it at all will see it always, those who don't will see it never. The trick will be figuring out how fix it for everyone without breaking it for anyone.
In reply to If it helps any, I have a… by mike320
FWIW, I use it heavily every day, exchanging score details with people ("suppose I changed measure 17 to be like this" etc.).
OK, something else to verify - and please, even people who thought they couldn't reproduce, please check - do you see different results form Ctrl+C (or "Copy" in menu) vs "Save As"? I do. The "Save As" commands (both of them) work fine, only "Copy" produces the issue.
In reply to OK, something else to verify… by Marc Sabatella
Confirming that both SaveAs's produce correct images. Copy (command and cmd-C) is a huge boon, though....
Sure. I can see right where the problem is, we don't do a re-calculation for MScore::pixelRatio on copy as we do for save as. The problem is there are different possible calculations here (PDF vs SVG vs PNG) and frankly it's going to be a bit of trial and error to get it the calculation right everyone on the "copy" operation, which is different still.
Maybe it's more clear here, I copied an image via image capturing and inserted it inside the score and used again the image capturing with "save as" (So yes, I confirm):
BTW: I saw this too today, snapshot mode and copy/paste it, but not when saving it
Seems to have come up again in https://musescore.org/en/node/280371
I can reproduce this issue on Linux, and my tests show that assigning correct pixel ratio makes the situation better but not perfect. There is also an inconsistency between DPI-related values assigned to the used
QImage
and topixelRatio
. I made a pull request that should resolve this (at least resolves the issue on my system), see https://github.com/musescore/MuseScore/pull/4490.Fixed in branch master, commit fc160b7947
fix #279377: unify PNG export and image copying to clipboard
Fixed in branch master, commit 060674dad4
Merge pull request #4490 from dmitrio95/279377-img-capture
fix #279377: unify PNG export and image copying to clipboard
Correction...seems to work properly on Win 10 now, thanks.
Automatically closed -- issue fixed for 2 weeks with no activity.