Stitching Plugin Failure

I have a problem with the Stitching Plugin and inconsistency with the Bioformats importer Plugin:
Basically channel 3 of a 4-channel Tile czi picture is lost in the stitching.
When Importing and stitching the picture with the drag and drop Bioformats:

Take into considerations the colors are changed from the original, but this dotty points are what should be seen in channel 3.
On the other hand, while stitching the pictures with the Grid/Collection Stitching plugin using:

  • Type: Positions from file
  • Order: Define by image metadata

Channel 3 is incorrectly imported, lost and fused with channel 4, or better said, shows the same thing than channel 4.

  • OS: Windows XP 32 bit
  • Fiji version: Windows 32 bit version downloaded 12.11.15

Thanks in advance!

To me it looks as if there is something missing here, no?

From the screenshot it’s pretty hard to guess what you’re exactly referring to. Could you post a cropped version showing the details of your statement?

To reproduce, a sample dataset would be helpful if you can post one.

Dear ehrenfeu,

I couldn’t put more pictures because of being a “new” member of the forum. Therefore I can only put 1 unfortunately, whichs is why my explanation is impossible to comprehend.

Trying to put it easily, the stitching plugin makes me loose a channel, specifically channel 3. I´m guessing it´s lost bound to one of the others, maybe channel 1…

I still couldn’t solve this issue. Is there anyone out there who can help me?

Thanks!

@Irupe I increased your account level from “new user” to “basic user”, this should allow you to post more than one image. Could you try to edit your original post and add the missing images/screenshots?

One of the admins (@ctrueden, @hinerm, @dietzc) might be able to increase the uploading restriction for new users (the ImageJ forum would be nothing without images :smile:)


If I understand correctly, you are seeing different image data when using Bio-Formats compared to when using the Stitching plugin? It might also be helpful if you upload the czi dataset in question via Help > Upload Sample Image. That way, the Fiji developers can have a look at it.

2 Likes

I bumped it up to 5 (arbitrarily) since this has come up a couple times now. I don’t think I’ve seen any spam on the forum yet, so I, personally, am not opposed to further relaxing the limits on new user permissions…

3 Likes

Dear all, I past it again as a reply because I didn´t manage to edit. Here it goes:

I have a problem with the Stitching Plugin and inconsistency with the Bioformats importer Plugin:
Basically channel 3 of a 4-channel Tile czi picture is lost in the stitching.
When Importing and stitching the picture with the drag and drop Bioformats:

Take into considerations the colors are changed from the original, but this dotty points are what should be seen in channel 3.

Channel 4:

On the other hand, while stitching the pictures with the Grid/Collection Stitching plugin using:
Plugins
Stitching
Grid/Collection Stitching

Type: Positions from file
Order: Define by image metadata

Channel 3 is incorrectly imported, lost and fused with channel 4, or better said, shows the same thing than channel 4 (my guess)
Channel 3:

Channel 4:

OS: Windows XP 32 bit
Fiji version: Windows 32 bit version downloaded 12.11.15

I tried the same file /s using the Fiji version of Windows 64 bit and it works perfectly, so it seems to be an issue for the Win 32 version of Fiji.
Thanks in advance!

Ah, I guess the confusion is with the “Stitching” option in the Bio-Formats importer: this one only “glues” the tiles together, it doesn’t do any registration. This is why there is the “Grid/Collection Stitching” plugin.

If you hover your mouse over the “Stiching” option in the importer, you will see a description of this on the right hand side panel, stating exactly this.

Does this answer the first part of your question?

1 Like

@ehrenfeu I think the issue is that channels 3 and 4 contain the same signal after stitching, whereas they are different in the raw data, as they are correctly shown using Bio-Formats only (without stitching).

@Irupe That’s indeed a strange bug that you’re seeing here. However, I wasn’t able to reproduce this with a czi file of my own and a 32-bit Fiji on Windows 7. Any chance that you can make your dataset available for testing?

1 Like

@imagejan agreed, that’s the second part of the question - at least how I understood it. I don’t have any 32 bit systems around any more (and I don’t see a purpose for them, especially in image processing), so no testing here… :wink:

Yay, I’m glad to know that someone reads those helps screens! :wink:

1 Like