ValueError: images must be same shape

[josenimo]

Dear ASHLAR devs,

I am trying to run ashlar on tiles imaged from MACSima instrument, and for some reason I am not able to run it. This was the command I used:
ashlar --output ./output.tif --align-channel 0 "fileseries|.|pattern=Cycle_{cycle:3}_Channel_{channel:1}_Frame_{series:1}.tif|overlap=0.075|pixel_size=0.65|width=2|height=2|layout=snake"

and the error I got

Stitching and registering input images Cycle 0: reading fileseries|.|pattern=Cycle_{cycle:3}_Channel_{channel:1}_Frame_{series:1}.tif|overlap=0.075|pixel_size=0.65|width=2|height=2|layout=snake Traceback (most recent call last): File "C:\Users\jnimoca\Jose_BI\Anaconda\envs\ashlar\lib\runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "C:\Users\jnimoca\Jose_BI\Anaconda\envs\ashlar\lib\runpy.py", line 86, in _run_code exec(code, run_globals) File "C:\Users\jnimoca\Jose_BI\Anaconda\envs\ashlar\Scripts\ashlar.exe\__main__.py", line 7, in File "C:\Users\jnimoca\Jose_BI\Anaconda\envs\ashlar\lib\site-packages\ashlar\scripts\ashlar.py", line 212, in main return process_single( File "C:\Users\jnimoca\Jose_BI\Anaconda\envs\ashlar\lib\site-packages\ashlar\scripts\ashlar.py", line 243, in process_single edge_aligner.run() File "C:\Users\jnimoca\Jose_BI\Anaconda\envs\ashlar\lib\site-packages\ashlar\reg.py", line 484, in run self.compute_threshold() File "C:\Users\jnimoca\Jose_BI\Anaconda\envs\ashlar\lib\site-packages\ashlar\reg.py", line 584, in compute_threshold _, errors[i] = utils.register(img1, img2, self.filter_sigma, upsample=1) File "C:\Users\jnimoca\Jose_BI\Anaconda\envs\ashlar\lib\site-packages\ashlar\utils.py", line 23, in register shift = skimage.registration.phase_cross_correlation( File "C:\Users\jnimoca\Jose_BI\Anaconda\envs\ashlar\lib\site-packages\skimage\registration\_phase_cross_correlation.py", line 225, in phase_cross_correlation raise ValueError("images must be same shape") ValueError: images must be same shape

conda list:
ashlar_conda_list.txt

Here you can see my test files, and the jupyter notebook that I used to regex the filenames.
https://filetransfer.mdc-berlin.de/?u=8Ek6pcgQ&p=7Xx3qkMZ

It is a simple 4 tile ROI. However it does follow a different imaging path (see image):

Thanks in advance for the help.
Best,
Jose

[Yu-AnChen]

Thanks for the detailed report and test images! I've filed a PR (#161) that fixes the error you are seeing.

Note that your image layout requires adding direction=vertical to the ashlar call. I used the following call and generated the attached screenshot

ashlar --output ./output.ome.tif --align-channel 0 "fileseries|.|pattern=Cycle_{cycle:3}_Channel_{channel:1}_Frame_{series:1}.tif|overlap=0.075|pixel_size=0.65|width=2|height=2|layout=snake|direction=vertical"

[josenimo]

Dear @Yu-AnChen,
thank you for the help, the direction=vertical was perfect.
I needed to downgrade to Ashlar 1.14.0 to not get the error.
Best,
Jose

[niovl]

Hi, I got the same error on my data. The data is like this,

script is like this，
ashlar "fileseries|D:/data/large_chip/A6/test|pattern={series}.tif|width=3|height=1|overlap=0.1" --align-channel 0 -o output.tif

[josenimo]

@niovl have you tried passing the layout argument? In your case it would be "raster"

[niovl]

I tried it with python, got the same same error，
reader = fileseries.FileSeriesReader(filespath, pattern='{series}.tif', overlap=0.11, width=3, height=1, layout='raster', direction='horizontal', pixel_size=1, )

ea_args = aligner_args.copy()
edge_aligner = reg.EdgeAligner(reader, **ea_args)
edge_aligner.run()

Tried update the command like this：
ashlar "fileseries|D:/data/large_chip/A6/test|pattern={series}.tif|width=3|height=1|overlap=0.1|layout=raster" --align-channel 0 -o output.tif

[josenimo]

@niovl hmm I think I cant help you, I would have tried the same thing. I can try replicating the error. Could you upload the three images? What was your installation like? Also, what is the error message?

[niovl]

Hi, @josenimo

The error message

File "D:\apps\anaconda\setup\envs\ashlar\lib\runpy.py", line 194, in _run_module_as_main return _run_code(code, main_globals, None, File "D:\apps\anaconda\setup\envs\ashlar\lib\runpy.py", line 87, in _run_code exec(code, run_globals) File "D:\apps\anaconda\setup\envs\ashlar\Scripts\ashlar.exe\__main__.py", line 7, in File "D:\apps\anaconda\setup\envs\ashlar\lib\site-packages\ashlar\scripts\ashlar.py", line 212, in main return process_single( File "D:\apps\anaconda\setup\envs\ashlar\lib\site-packages\ashlar\scripts\ashlar.py", line 243, in process_single edge_aligner.run() File "D:\apps\anaconda\setup\envs\ashlar\lib\site-packages\ashlar\reg.py", line 484, in run self.compute_threshold() File "D:\apps\anaconda\setup\envs\ashlar\lib\site-packages\ashlar\reg.py", line 584, in compute_threshold _, errors[i] = utils.register(img1, img2, self.filter_sigma, upsample=1) File "D:\apps\anaconda\setup\envs\ashlar\lib\site-packages\ashlar\utils.py", line 23, in register shift = skimage.registration.phase_cross_correlation( File "D:\apps\anaconda\setup\envs\ashlar\lib\site-packages\skimage\registration\_phase_cross_correlation.py", line 225, in phase_cross_correlation raise ValueError("images must be same shape") ValueError: images must be same shape

The three images：
test.zip

[niovl]

Hi， @Yu-AnChen ，Hope to get your help.