Quote from: Svampen on January 19, 2012, 09:47 AM

I ran into memory issues

Please provide some specifics on your system and the Windows Virtual memory/swap-file settings

Well it was an old laptop with like 1GB of memory. I would have split the capture up anyways, so would be awesome to have it done automatically.

Hit guess margins, THEN manually tweak them, THEN hit "guess overlaps"

Sounds like what I did, but I left the scrollbars in place to be able to find the stitch-seams

One suggested modification to your workflow:

Hit guess margins, THEN manually tweak them, THEN hit "guess overlaps"

Overlap guessing is based on margins, so tweaking them manually before guessing overlaps is advantageous.

Thanks for the tip -- hope to try that soon

Ok it's time for me to roll the scrolling capture code into Screenshot Captor itself.. Now i just have to decide how to deal with non-scrolling object capture..

In current SC, when you want to capture an object or scrolling object, you trigger object captor and then click differently if you want a scrolling capture.

I could keep that the same way, or i could have it pop up a dialog as soon as you say you want to do an object/scrolling capture asking which type.. But that could slow down the workflow.  On the other hand, I dont want to make it separate functions triggered by separate hotkeys..

I could have a multi-step process where first you select an object/region using a single method -- and then get asked whether you want to: 1) just copy the image of the object, or 2) begin a scrolling capture of the object, or 3) capture text of object.

^+1

Why not keep the current selection method, and replace the old Scrolling capture with the New And Improved! Scrolling capture? Should leave any previous workflow the same, imho.

Yeah i guess that's mostly what i was suggesting.. The only change would be like so:

OLD METHOD:

Trigger an object/scrolling capture, see red box, move red box to where you want it, click for object capture, or shift+ctrl+click for scrolling capture, or menu for text capture.

NEW PROPOSED METHOD:

Trigger an object/scrolling capture, see red box, move red box to where you want it, click to select.  Then get presented with a dialog with your 3 choices of copy object/region previously selected, or do a scrolling capture of the region previously selected, or do a text capture of the object selected.

ANOTHER POSSIBILITY:

Trigger an object/scrolling capture, then get presented with a dialog with your 3 choices of copy object/region, or do a scrolling capture , or do a text capture, then get shown red box to position.

Ath, that is a good idea, but my first instinct is to minimize the options and features of that red box window that follows the cursor around, just because i worry about it being confusing.  Let's see if my new version is adequate.

New beta of Screenshot Captor is up that uses the new Scrolling Capture system:

http://www.donationcoder.com/Software/Mouser/screenshotcaptor/downloads/beta/ScreenshotCaptorSetup.exe

The new system is also used to capture "objects" even when you don't want to do scrolling.

In order to make workflow efficient, instead of the fun introductory dialog that was shown in the first scrolling capture implementation -- now you get the red box right away, and after you select an object you get a dialog.



One improvement in workflow is that it's now easier to understand how to do multiple OBJECT captures in a row -- useful if you want to grab various regions of a window -- and there is natural place for me to add features on window objects -- like adding an option to "capture all children windows", on the post-object-selection dialog.

Let me know what you think!

like adding an option to "capture all children windows/controls"

I know this is something people have asked for before.. i wonder if there are people still want this?

New version uploaded; the previous one broke the scrolling capture margin and overlap detection functionality.

In order to make workflow efficient, instead of the fun introductory dialog that was shown in the first scrolling capture implementation -- now you get the red box right away, and after you select an object you get a dialog.
(see attachment in previous post)

Let me know what you think!

Mouser, man!, you did a great job! I like it this way

The dialog is a bit intimidating, larger fonts on the buttons and all, but it's pretty useful.
Be aware that you can never have enough explanation-text on that dialog, as it's IMHO quite complicated for new users. A big tooltip for each button (and enough tooltip-time to read) or a specific help link or pop-up button next to each could be needed.

Mouser-ish eyes only

Just found another though cookie to scroll-crack: Visual Studio 2010. It's WPF based to start with, and when scrolling it tries to keep the last line as the first line of the next page

I did an 'autodetect' for the scrolling method, and the default seemed just fine, setting the number of pages correct (after a test-scroll) does definitely improve the steps of guessing margins and overlap, but I can't get it to do the actual overlap correct: the overlap is not correctly guessed, and manually changing something only corrects the one that you're looking at.
I made a few snippets after making a code screenshot:
Seam of page 1 and 2:

The seam is a bit too wide, the red boxes are copies of the same size, the green box is about the diff
Seam of page 2 and 3

There is some overlap to be resolved, as there is actually only one line saying that, not 2
Seam of page 3 and 4

More of the overlap is resolved, but still no sigar.

If I manually adjust so the third overlap is correct, only about half of the second overlap is covered. The first overlap is quite hard to view/correct.

Tomorrow or Monday (don't know when/if I have enough time to be playing with this) at work I'll try to scroll-capture another WPF application, to see if that is working somewhat better.