Image depth(?) issue, through Beta 3

[mrbroschat]

Evolution of the issue:

I used Paint.NET (3, beta 2) to scan let's say ten pages from a book. Then I fiddled these in various ways to make the eventual OCR process more successful (mostly, sharpening). Then I used the new Adobe Acrobat 8 to make a PDF collection of the set. Each image in the resulting PDF collation was nonsense, unrecognizable. I wrote this off as a probable inability of Acrobat to handle PNG (my format of choice), and went on to other things.

After similar work involving more pages, a lot of OCR, different kinds of images, etc, which I had done with a mixture of Photoshop Elements (PSE) and Paint.NET (PdN), I again created a PDF with the set of images. Perfect. Except for one page, which was unrecognizable. On a hunch, I opened it in PS Elements, and saw that although I had scanned the pages as greyscale with 300-bit resolution, PSE was seeing it as an RGB image (PSE has a command called Image, which includes Mode: Bitmap, Greyscale, Indexed Color, RGB Color; I don't see any equivalent in PdN). I clicked the mode to Greyscale (generating no visible change in the image), and tried the Acrobat PDF creation once again. Perfect.

So, PdN does something to those images. Is it that Acrobat can't handle RGB (together with greyscale)? Nope. The first image in the set is a big RGB image, while all the remaining pages start out as greyscale. All images were scanned at 300 dpi.

PdN doesn't have (or I haven't found) anything that reports RGB, greyscale, etc (or bit depth or resolution), and it was my normal viewer (IrfanView) that showed the difference in bit depth between what proved to be the problem PdN image and the rest (which had started in PdN but then been manipulated in PSE).

Having written all this, I suppose it's possible that Acrobat simply cannot handle whatever PdN generates as a 24-bit image (if the successful 24-bit image was created in PSE), but then why did PdN transform a PSE 8-bit greyscale image into 24 bits in the first place?

Am more than a bit confused about what's going on, but because we're still in beta, I thought I'd throw out my experience. The last steps of my experience were done with Beta 3, so I'm guessing that whatever I've seen still exists.

Broschat

[mrbroschat]

I still don't know all the issues, but here at work I've scanned a page of text (using the greyscale option in my scanner driver), saved the image as PNG, and directly made a PDF using Acrobat 7. Works perfectly.

Yes, Paint.NET does generate a 24-bit file from this 8-bit input, but that in itself doesn't affect Acrobat's ability to deal with the image. Afterwards, I used OptiPNG to optimize the PNG image, and the image was reduced nearly in half, converting to 8-bit in the process. That, too, went into Acrobat without difficulty.

Will re-test with Acrobat 8, but am no longer concerned about any contribution from Paint.NET to the problems I had seen...