Software for a simple 2D panorama stitch from 3 PI-XAS scans

[grat]

My bad on the perspective mode-- as I said, I don't rent lightroom, and was looking at a much older page apparently.

Please post your individual frames and a sample of the distorted final image. saying "it don't work" and "there is no distortion" can only go so far in allowing us to help.

 

[Les Sarile]

My bad on the perspective mode-- as I said, I don't rent lightroom, and was looking at a much older page apparently.

Please post your individual frames and a sample of the distorted final image. saying "it don't work" and "there is no distortion" can only go so far in allowing us to help.

I'm certain I can do this manually let alone fully automatic programs like MS ICE having any problems at all doing this.
This example of 3 frames of Kodak GC400 was the hardest one I've had to stitch using MS ICE as it was taken with a wide angle lens so I needed to adjust each frame using perspective first. Not perfect but not too distracting.

Kodak GC400-8_01-17-18-191 by Les DMess, on Flickr

Otherwise, I've done MS ICE stitches like this one taken portrait mode of 9 frames of Kodak Ektar 100 automatically.

Kodak Ektar 100_31-16to24_stitch C by Les DMess, on Flickr

Definitely curious what the difficulty is stitching essentially flat continuous scans of the same frame.

 

[brbo]

Definitely curious what the difficulty is stitching essentially flat continuous scans of the same frame.

Well, why don't you try it? Stitch a panoramic frame with different stitching modes and you will see that all resulting panoramic pictures will be different. Obviously, OP wants to know which mode is the best (being the closest geometrically to a picture that was scanned on a scanner that can scan the entire panoramic frame in one go).

 

[Radost]

Well, why don't you try it? Stitch a panoramic frame with different stitching modes and you will see that all resulting panoramic pictures will be different. Obviously, OP wants to know which mode is the best (being the closest geometrically to a picture that was scanned on a scanner that can scan the entire panoramic frame in one go).

Exactly. I was hoping there is a simple tool that preserves the geometry while stitching.

 

[Les Sarile]

Exactly. I was hoping there is a simple tool that preserves the geometry while stitching.

Are you saying the geometry is not perfect from multiple flat scans of one frame of film? So much so that even the mildest of perspective correction cannot make it straight? Curiouser and curiouser indeed . . .

 

[Radost]

Are you saying the geometry is not perfect from multiple flat scans of one frame of film? So much so that the mildest of perspective correction cannot make it straight? Curiouser and curiouser indeed . . .

I don't want perspective correction. I shoot with minimal distortion lenses and scan with XAS scanner. But I dont want the software stitching to change the geometry of my 3 scan panorama.
I am in a middle of a war zone at the moment and can not overlay layers in photoshop to check the acuracy of photomerge in photoshop. But Lightroom panorama defenetally introduces distortions.

 

[Les Sarile]

I don't want perspective correction. I shoot with minimal distortion lenses and scan with XAS scanner. But I dont want the software stitching to change the geometry of my 3 scan panorama.
I am in a middle of a war zone at the moment and can not overlay layers in photoshop to check the acuracy of photomerge in photoshop. But Lightroom panorama defenetally introduces distortions.

First of all, I wish you the best given your situation. That concern obviously negates all others!

If I understand what you did, you have one long piece of film that was scanned in three parts. I anticipate each one to be perfectly flat and square - no additional perspective issues introduced in the scans. Hence no perspective corrections needed. In this case, I expect to simply overlay the scans and have no perspective issues introduced at all. Given the nature of scans, I can expect some possible differences in brightness or perhaps even color that may be apparent in the overlap areas. Another way to address color/brightness is during the scan if your software has the feature of disabling autoexposure so that all three scans will be perfectly the same in that regard. If not, MS ICE also compensates for these exposure differences as well.

Out of curiosity, what is the size of that film?

 

[Radost]

First of all, I wish you the best given your situation. That concern obviously negates all others!

If I understand what you did, you have one long piece of film that was scanned in three parts. I anticipate each one to be perfectly flat and square - no additional perspective issues introduced in the scans. Hence no perspective corrections needed. In this case, I expect to simply overlay the scans and have no perspective issues introduced at all. Given the nature of scans, I can expect some possible differences in brightness or perhaps even color that may be apparent in the overlap areas. Another way to address color/brightness is during the scan if your software has the feature of disabling autoexposure so that all three scans will be perfectly the same in that regard. If not, MS ICE also compensates for these exposure differences as well.

Out of curiosity, what is the size of that film?

Thank You. I am good at the moment.

I am using 135 film inside Fuji GSW690III 65MM F5.6. I scan it RAW with a lot of overlap on a pacific image scanner so no brigthness or contrast difference when I stitch. Just the overcomplicated panorama stitching on all programs I used.
I will try PhotoMerge in photoshop when I get the chance. I will even try to just manually align the picture in photoshop.
I was hoping to get some simple cheap tool so I dont have to pay for lightroom and photoshop.

​

 

[Les Sarile]

Thank You. I am good at the moment.

I am using 135 film inside Fuji GSW690III 65MM F5.6. I scan it RAW with a lot of overlap on a pacific image scanner so no brigthness or contrast difference when I stitch. Just the overcomplicated panorama stitching on all programs I used.
I will try PhotoMerge in photoshop when I get the chance. I will even try to just manually align the picture in photoshop.
I was hoping to get some simple cheap tool so I dont have to pay for lightroom and photoshop.

​

You didn't say but did you try the MS ICE in the link I posted earlier? It's better than cheap . . .

 

[Radost]

You didn't say but did you try the MS ICE in the link I posted earlier? It's better than cheap . . .

Dont have a PC on the field. Will do in september when I go back to the states.

 

[brbo]

I don't seem to you have much of a choice, either you align the scans by hand or use proper tool with proper settings for the job. I've shown you the example stitch with proof that no geometry correction is done in PS Photomerge with reposition mode. MS ICE will be more like 'collage' mode.

Both, reposition and collage mode don't introduce geometry correction per se, but they differ in how the joints between partial scans will be made. The 'collage' mode will rotate (this is not geometry altering of the partial scan) the partial images to create the best joints when 'reposition' won't even rotate images (so the joints will be of slightly worse quality). In ideal situation, when you have partial scans made with absolutely zero longitudinal tilt of the film in scanner holder between the partial scans, scanner lens focused at exactly same position, film with exactly same arch because of scanner induced heat...., both will produce the same result. Of course, none of this will be true in real life.

This is an exaggerated case of what you would get if you had a huge difference in film alignment between partial scans:

Left & right part:




PS Photomerge 'reposition':



PS Photomerge 'collage':



MS ICE:

 

[MattKing]

It seems to me that there is a real possibility that the geometric distortion observed by Radost would be visible in a single contact print of the film. The challenge being experienced is that Radost is attempting to straighten something that isn't straight in the film in the first place.

 

[Radost]

It seems to me that there is a real possibility that the geometric distortion observed by Radost would be visible in a single contact print of the film. The challenge being experienced is that Radost is attempting to straighten something that isn't straight in the film in the first place.

Nope. My horizon is strength in most of the pictures. Just don’t want the stitching software to distort.
Does noritsu scan panorama. It might be easier to just send my panorama pictures to Richard’s.

 

[MattKing]

I'm not thinking about the straightness of the horizon, which quite likely would be straight if it is anywhere near the centre line.
I'm thinking about:
1) either barrel or pincushion distortion; and
2) the fact that the ends are farther away from the camera than the middle, and therefore smaller on the film.

 

[Les Sarile]

Nope. My horizon is strength in most of the pictures. Just don’t want the stitching software to distort.
Does noritsu scan panorama. It might be easier to just send my panorama pictures to Richard’s.

A Noritsu scanner can scan a full 6X9 film so it can scan your film in one pass.

 

[brbo]

A Noritsu scanner can scan a full 6X9 film so it can scan your film in one pass.

Noritsu can scan 6x12 (if you purchase additional software extension from Noritsu) on 120 film. But Radost has 135 film and you can’t feed that into the MF carrier without a lot more trouble than a simple stitch. Frontier SP-2000/3000 manual MF carrier should be a fair bit easier to do that, imho.

Anyway, I’ve shown here how to do it properly with absolutely no distortion, but some people just can’t be helped…

 

[Radost]

Noritsu can scan 6x12 (if you purchase additional software extension from Noritsu) on 120 film. But Radost has 135 film and you can’t feed that into the MF carrier without a lot more trouble than a simple stitch. Frontier SP-2000/3000 manual MF carrier should be a fair bit easier to do that, imho.

Anyway, I’ve shown here how to do it properly with absolutely no distortion, but some people just can’t be helped…

As I said I will try your way when I leave the war zone.

 

[grat]

Obviously, OP wants to know which mode is the best (being the closest geometrically to a picture that was scanned on a scanner that can scan the entire panoramic frame in one go).

No, OP is claiming no mode works-- every mode is producing distorted images.

Exactly. I was hoping there is a simple tool that preserves the geometry while stitching.

And in my experience, most of them do. You might need to crop the final image for instance, if you shot a panorama by hand, but stitching multiple scans of one image, under controlled circumstances, should be producing the trivial use case.

The challenge being experienced is that Radost is attempting to straighten something that isn't straight in the film in the first place.

We don't know what the challenge being experienced is, because without being able to see the problem, we're all guessing.

 

[brbo]

No, OP is claiming no mode works-- every mode is producing distorted images.

In software that he used (or was able to figure out how to use it). Lightroom and Hugin (afaik, haven't worked with it much), indeed, can't do what he wants to do.

 

[grat]

In software that he used (or was able to figure out how to use it). Lightroom and Hugin (afaik, haven't worked with it much), indeed, can't do what he wants to do.

I haven't used lightroom in a really, really long time (version 1.0), but the idea that it can't stitch flat images together without creating distortion is just nonsensical. Similarly, Hugin (which the OP has difficulty with), while complex, can correct just about any possible distortion. The resulting image may have bits and pieces sticking out, but can certainly be cropped to a rectangular image free of (most) distortion.

Either the software is being used incorrectly, or the source images are providing their own distortion.

 

[brbo]

You, obviously, still don't get it. It's not about stitching software being unable to correct distortion (of any kind).

We've established which software (even freely available if using Windows is possible) can do it. But if you want to further educate OP, you can take two files (here and here) and show OP how stitching them together with Hugin is done so that no pixel is skewed.

 

[grat]

Well, here's the simple version processed entirely in Affinity, including inversion.



Stitching was literally "create panorama", add both files, hit "stitch".

While the film frame itself is curved, note that the image itself is straight, and there are no distortions in the line of the roof edge, for instance. I left the register in deliberately, but cropped down to just the image, there's no visible distortion.

Inversion was slightly more complicated. Most of the time was spent correcting the overall scene, as it seemed a bit washed out to me.

My Hugin install (linux) seems to be a bit confused right now-- if I can get it straightened out (so to speak), I'll post a sample of that too.

 

[Radost]

I haven't used lightroom in a really, really long time (version 1.0), but the idea that it can't stitch flat images together without creating distortion is just nonsensical. Similarly, Hugin (which the OP has difficulty with), while complex, can correct just about any possible distortion. The resulting image may have bits and pieces sticking out, but can certainly be cropped to a rectangular image free of (most) distortion.

Either the software is being used incorrectly, or the source images are providing their own distortion.

I don’t want it to correct distortion!!!!
I don’t want them to change the geometric integrity of my photos.

 

[brbo]

While the film frame itself is curved...

... and the rest, too. You may not care about that since the distortion is not huge, but some people would rather have the scan exactly the same as the negative. It's not any harder, you just need to use a proper tool.

 

[grat]

I don’t want it to correct distortion!!!!
I don’t want them to change the geometric integrity of my photos.

Sorry-- implied was that if it can correct distortion, a non-distorted image should result in zero distortion.

But it's still not entirely clear, at least to me, what type of distortion you're experiencing.