Version 0.9.9pre4; lots of changes; needs testing

Scan Tailor specific announcements, releases, workflows, tips, etc. NO FEATURE REQUESTS IN THIS FORUM, please.

Moderator: peterZ

User avatar
Misty
Posts: 481
Joined: 06 Nov 2009, 12:20
Number of books owned: 0
Location: Frozen Wasteland

Re: Version 0.9.9pre4; lots of changes; needs testing

Post by Misty »

Tim wrote:Yes, you can clear an image zone you added, but you can't clear any of the ones that were identified automatically.
No, you actually can subtract from an automatically identified zone - unless that changed from 0.9.8. You can't clear it by right-clicking, but you can define a zone which subtracts from the auto zone instead of adding to it. Like I said, create an image zone and right-click on it. There is an option to make your zone subtract from the auto zone instead of defining an additional image zone.

Edit: Here's an example of what I'm talking about. As you can see, the illustration was autodetected, and I drew a zone which is subtracting from the blue autodetection zone.

(Please note that this page segment is from a copyrighted work and should not be redistributed.)
Attachments
subtract.png
subtract.png (184.08 KiB) Viewed 7855 times
The opinions expressed in this post are my own and do not necessarily represent those of the Canadian Museum for Human Rights.
Tim

Re: Version 0.9.9pre4; lots of changes; needs testing

Post by Tim »

Now I get it, thank you. I hadn't thought of going into the properties menu and trying that. Yes, that still works in 0.9.9pre4, and accomplishes what I would like to be able to do. Not quite as easy as right clicking to clear, but it works.
Tulon
Posts: 687
Joined: 03 Oct 2009, 06:13
Number of books owned: 0
Location: London, UK
Contact:

Re: Version 0.9.9pre4; lots of changes; needs testing

Post by Tulon »

strider1551 wrote:I pulled the latest from git last night. In the fix orientation stage, apply to every other page wasn't functioning.
That was a good, recently introduced bug. May it rest in peace ...
Scan Tailor experimental doesn't output 96 DPI images. It's just what your software shows when DPI information is missing. Usually what you get is input DPI times the resolution enhancement factor.
Tulon
Posts: 687
Joined: 03 Oct 2009, 06:13
Number of books owned: 0
Location: London, UK
Contact:

Re: Version 0.9.9pre4; lots of changes; needs testing

Post by Tulon »

Tim wrote:The thing I have noticed is that the Color/Grayscale mode in the output stage "forgets" about all the content selection and page layout from the previous steps and outputs an image to disk without any of those steps being done.
It doesn't "forget" page layout. It respects its geometry, but it doesn't clear the margins, unless explicitly told to do so.
Am I correct to assume that the Color/Grayscale choice doesn't do any image processing to the image (such as despeckling, etc) if the equalize illumination option isn't chosen?
It only does geometric transformations in this case.
In other words, do I get an non-degraded image as output if I choose no despeckling? If so, would it make sense not to change the DPI of the results by default? If not, what processing is being done in the output stage.
You will get a geometrically transformed image anyway. Making output DPI match the original would theoretically give you benefits only if no deskew was applied. In practice, you won't get any benefit even in this case, because ST doesn't even try to match input and output coordinates 1 to 1, knowing in most cases it's not possible anyway. Having said that, you don't need as high output DPI for grayscale images as you would for B/W ones.
The second thing I have noticed is the scrollbar in the preview/thumbnail pane doesn't allow scrolling to all images because the previews seemingly want to keep running up off the top of the page, makes work more difficult than it used to be.
I don't understand you here. Please be more specific.
I find I repeatedly have to hit page up to get back to the page I want to work on as I go through the stages. It would help if each stage reset back to the first image instead of starting on the last. I see that some of the changes were intended, but with no scrollbar *or* way to scroll with the mouse it makes it harder.
Making batch processing stop at the last page, or jump back to the top, or jump there when you launch it the next time - all of these options have pros and cons. Before the next release I'll probably make up my mind on which one to adopt.
I also see that the changes to the Page Split stage were intended, and the new code is much more accurate in picking the correct side, but I found the green arrows useful to switch if needed.
There are no more "sides" when dealing with single page scans / shots. We now have two cutters and the page area is what between them. No sides means no need to switch sides.
Scan Tailor experimental doesn't output 96 DPI images. It's just what your software shows when DPI information is missing. Usually what you get is input DPI times the resolution enhancement factor.
Tim

Re: Version 0.9.9pre4; lots of changes; needs testing

Post by Tim »

Ok, the forgetting part and transformations part makes sense. I didn't notice the forgetting really was just a difference of the margins.
Tulon wrote:
I don't understand you here. Please be more specific.
Ok, maybe I can do better this time. Now after each stage, the preview pane's scrollbar no longer covers all the images in the project. Some have been "lost" off the top, and the only way to get back to all of them is to hit pg up once for every image in the project until you're back to the first image. This happens after every stage as far as I can tell. At various other times the scrollbar gets lost as well and again the only way I can find to get back to previous images is to hit pg up.
Tulon
Posts: 687
Joined: 03 Oct 2009, 06:13
Number of books owned: 0
Location: London, UK
Contact:

Re: Version 0.9.9pre4; lots of changes; needs testing

Post by Tulon »

Tim wrote:Now after each stage, the preview pane's scrollbar no longer covers all the images in the project. Some have been "lost" off the top, and the only way to get back to all of them is to hit pg up once for every image in the project until you're back to the first image. This happens after every stage as far as I can tell. At various other times the scrollbar gets lost as well and again the only way I can find to get back to previous images is to hit pg up.
I can't reproduce that. Maybe you could make a screencast to demonstrate the problem.
BTW, what exactly do you mean by "after every stage"?
Scan Tailor experimental doesn't output 96 DPI images. It's just what your software shows when DPI information is missing. Usually what you get is input DPI times the resolution enhancement factor.
Tim

Re: Version 0.9.9pre4; lots of changes; needs testing

Post by Tim »

Tulon wrote: I can't reproduce that. Maybe you could make a screencast to demonstrate the problem.
Yeah, it's a pretty big one, I assumed you didn't see it for one reason or another. I'll put together a screencast as soon as I can.
BTW, what exactly do you mean by "after every stage"?
I tend to like to go through each stage (Fix orientation, deskew, etc) one by one and see how the output from each looks before I proceed to the next. So after I've finished batch and/or manual processing for each image on a stage and then click on the name of the next stage would be precisely what I meant.
Tim

Re: Version 0.9.9pre4; lots of changes; needs testing

Post by Tim »

Ok, I discovered my desktop computer is too slow to make a screencast, the dropped frames ran about 77%. So I'll have to put together some screenshots to explain what I mean, or see if I can get one on my Mac. In trying to test scantailor on the Mac I did get a crash, but all it gave on the command line was 'Segmentation fault'. Apple's crash reporter intercepted the rest, and hopefully it is useful. I can send you the output if you think it would be.
Tulon
Posts: 687
Joined: 03 Oct 2009, 06:13
Number of books owned: 0
Location: London, UK
Contact:

Re: Version 0.9.9pre4; lots of changes; needs testing

Post by Tulon »

Tim wrote:Ok, I discovered my desktop computer is too slow to make a screencast, the dropped frames ran about 77%. So I'll have to put together some screenshots to explain what I mean, or see if I can get one on my Mac. In trying to test scantailor on the Mac I did get a crash, but all it gave on the command line was 'Segmentation fault'. Apple's crash reporter intercepted the rest, and hopefully it is useful. I can send you the output if you think it would be.
Yes, please send it.
Scan Tailor experimental doesn't output 96 DPI images. It's just what your software shows when DPI information is missing. Usually what you get is input DPI times the resolution enhancement factor.
Post Reply