DNG center crop for stitching

Hi all!
As it’s known fact that most lenses have marginally better center performance compared to corner performance, I would like to stitch DNG center parts only.

Unfortunately Lightroom panorama does not agree even if you have tons of images to choose from it always chooses to use some image corners which are un-sharp and does not perfectly match with other images as used in stitch source.

So far I have not found a way to truly crop DNG nor RAW files and I preferably would not like to convert files to tiff. I also would like to stitch before using NLP (I know that I can first convert all stitch source images individually in NLP, then crop unsharp corners and finally stitch final image) as it’s faster and more universal workflow.

Any tips would be appreciated!

If you happen to have a full frame camera and it happens to have a crop sensor mode, that might be one option.

How large are you going to print your photos?

I find that stitching helps to get the film grain sharper - and that output from a 24 Mpixel camera printed on common size papers looked at from reasonable distances does not push me to stitch. The situation might be different with a bad lens and a lower Mpixel camera though.

Make sure to apply lens corrections before stitching. It can help Lr to get a better result.

I look photos on screen and I can see the point where corner part is stitched to center part of the image, it’s easily visible on 800+ iso film grain. For very fine grain film (Portra 160, 160NS), this issue is very minimal and you have to pixel peep to see the transposition part from corner to center part.

I use Sigma 70mm macro with tubes to get about 2:1 magnification (this is the reason I have poor extreme corner performance, but stitching software does not care and still uses extreme corner parts and when they are side by side with center part, you can see the transition very easily as center part is very-very sharp next to extreme corner). For single 35mm frame I usually take 12 shots for stitching.

Up until now I first convert all stitch source images individually in NLP, then crop unsharp corners and finally stitch final image but ideally I would like to use NLP as last step to speed things up.

You might try overlapping your shots a bit more and use Photoshop to merge/pano instead of LR. Also, are you using a middle range f/stop say f/8 or f/11?

I do use Photoshop too and I have a quiet a bit of overlap but still time to time extreme corners are used. So there might be more solution how to get more even sharpens, like going for f/22-32, I try to aim maximum details I can extract from 35mm film and so far with the reasonably priced equipment I see that going for center crops with f/10-11 is the best solution. And this solution would be much easier if I could crop DNG files or force LR or Photoshop to only use center crops of stitching source.
One additional benefit for center crop is that your film does not have to be 100.000% in level with the sensor and its much easier to check the focus as you only need to check it from the center and you can pretty much ignore the corners.

I’d not recommend going down to the smallest f/stop as this will tend to introduce refraction issues. F/8 to f/16 would seem to be the best.

I’ve not tried it, but just off the top of my head, use layers in PS and then select and mask the area you want, then try doing the pano merge. Just an idea off the top of my head.