Using different lenses?

hello everyone.

I was wondering if it was possible to use different camera lenses to scan a 3D model?

for example a wide one (25mm) for the global structure and a standard or zoom (50-80mm) for the details?

How RC will threat those info? Can it make the difference between the two and use the images from the 20mm to build the mesh and the images from the 50-80mm to create the textures automatically, for example?

What is RC limitation about that?


It doesn’t matter how many lenses (even cameras) as long as you take more than just a few images with one focal length.

You can look around in the forum (use search) for more questions on this subject…  :wink:

yeah i normally use a mix of lenses, anything from 100mm down to 14mm.

I’ve done some tests with 10.5mm which considering it doesn’t support fisheye’s yet. i still got some results from it.

By default RC treat all images as independent camera/lens and estimate all settings per image.
If you enable group by EXIF on load or group images any new focal lenght or even camera rotation treat as different EXIF lens group.
If you use one prime lens and removed all rotation tags and group images, RC can use information from all images for lens/camera settings estimation and result can be more precised.

But if you use different lenses, zoom, or even different cameras DSLR, mirrorless, smartphone RC will anyway work with them without problem. Just quality must be good.

chris, could you elaborate on your results with the fisheye?

well i was stitching them together with laserscan data.

and it worked well up to a point… until it stopped working.

I didn’t end up using any of them, and re-shot it with 14mm.

maybe it was when i tried to add normal photos. it was a little while ago now.

I could stitch with division on fisheye. and normal photos with brown 3.

then merging those together it wasn’t too happy.

I also did a few tests with 360 pano rig, and shooting 360’s with the fisheye, and sending in all the individual photos to rc.

but i found this didn’t work as well and shooting normally for photogrammetry.



Thank you!

What was the problem exactly? Misalignment? Differences in the geometry? Texture off?

Would be interesting to see a comparison…  :slight_smile:

Vlad Kuzmin:

I don’t get why we have to delete the rotation info in the EXIF? Can you help me out on this?

Also is it needed only when you have multiples lenses or is it a common thing to do first no matter what?

Enlighten up us there :slight_smile:


Ok. Lets imagine you have 50mm prime lens.
And try to imagine that your lens is not perfect. And have some distortion in top left corner. But this distortion still can be estimated with Brown or Brown+tangent lens model.

So all images in your dataset have exactly the same lens distortion model.
But camera add rotation tag depend on how you rotate camera.
And main problem that this tag can be not correct.
Or Photoshop will unrotate image but forgot delete tag, Or rotate your image to -90degree and delete tag and other to +90 degree and also remove tag…

So we can’t trust this tag so much.

But if you remove this tag from your RAWs right after copying from SD card. You will have all images horisontal with top left corner exactly the same on all images.

And main advantage, that all this images can be grouped in one lens group. So undistortion model will be estimated from all images and features detected on them. This mean more precise results.

But as soon as focus can give some small differences, before final refine Align better ungroup images and RC will finally adjust all undistrotion models per image.

Hey Mike,

if you’re just starting out, don’t worry too much about the exif stuff. This is only if you use automatic exif grouping and encounter problems. I used it with the rotation tag and never had issues. Also, if you have only a few lens settings and not too many images, you can easily do it by hand, which I also do at times. You will loose the auto rotation in RC, which was added not too long ago because if you have to add Control Points it can be quite a pain to do it on rotated and not-rotated images at the same time. So it’s a trade-off…

Vlad: deleting the rotation tag is needed when you have multiple lenses or it’s better to do it everytime? Also is it possible to take some photos in “landscape mode” and some in “portrait”?

And how do you “group” the pictures in RC?

thanks for your help btw :slight_smile:


Götz: yes i start but i want to know the more possible :slight_smile:


It took me about 13 seconds with search (“group images”) to find this:

I’m off now…


I can’t find any practical reason do not group cameras by default. Ungrouping on final refinement Align also on your own taste.

BTW, this is not required any additional works. Just enable grouping on image load in global settings.

Götz: oh sorry I should have looked it up! You’re nice, thank you!! :slight_smile:

Vlad: so it’s just a good habbit to do so. The thing I dont understand is when you talk about “taste”. As a technical workflow there is no taste: there’s a way it should work better (or at least help things up) and one that works less good.


is that workflow correct to you?


also can I take photo horizontaly and vertically? 

thank you for your time :slight_smile: I appreciate that

Hi Mike!

You must not care about camera rotation. Always try to shoot as much data from object as possible. If this will required vertical or horisontal or diagonal :wink:

Taste… more about experience, align result and align report. In most cases Ungroup and last one Align after this must increase quality. In some situations you can lost some cameras or have more weird effects…

Mike, don’t get me wrong. I like to help where I can, but there are just some basic things that won’t take too long to read through yourself before asking in a forum like this. The help in RC is not very good if you want to look up something specific, but it is well made for a first walkthrough and will answer most of the initial questions.

About the taste:

I totally agree with Vlad on this - photogrammetry, especially with RC, is more of an art than precise science. Yes, the results can be highly accurate, if you know what you’re doing. But to get there you need to build up your own experience, nobody can spare you that. The variables in terms of objects, techniques and aims are so big that basically every project is unique and requires it’s own workflow.

Vlad: I see, thank you!

Götz: Yes, I understand sorry for that! There are so many ways/options to do or not to do, I am trying to figure out what is THE BEST WORKFLOW. The one that gives the best results every time. And it doesn’t matter if it takes more time or for some case it’s not necessary to do some actions as long as I get the best result :slight_smile:

No need to apologize!  :slight_smile:

But definitely need to try around for yourself!!!

Believe it or not, I can totally understand your approach! I always want to know everything beforehand myself. Only I had to learn that it is not possible for this kind of work. There is no standard workflow! There might be one that works for you personally, but only after figuring it out over many projects… :slight_smile:

the full-text search for Help is being worked on :wink: