Peter has asked some interesting questions over here that I would like to address.
Here are some answers:
Were you only using your 50mm lens for this like you described in your tutuorial?
No, I was using my 70-200mm lens at 200mm, because I had to zoom in quite close. The result is still a wideangle (large Field of View), because there are quite a few pictures stitched together. One of the problems you will face though is the compression in depth of field a telephoto lens will cause.
When you are taking the pictures do you ever overlap so there will be 3 pictures with the same stitch joint in it?
The pictures overlap by about 20% in either direction. This is necessary, otherwise you couldn't make sure you won't have holes (the images get slightly warped to fit) and since it would be impossible to line them up exactly, plus the software needs the overlap area to figure out how the images go together.
join them in autopano it makes it rather soft
All Autopano does is figure out the control points (where the images overlap). The actual blending is done by either Enblend or Smartblend.
In PTAssembler you should turn off Feather (marked red in this part of the tutorial) plus you might want to set the Interpolator to Sinc1024.
do you usually just use the picture right out of the program (cropped, ofcourse)
I use the pictures without any adjustments, but don't crop them either. You will just crop away the information Autopano and Enblend so desperately need.
finally, do you always force it in manual exposure or do you have success with joining files that are auto metere?. just because i would think sometimes the whole scene is too constrasy to keep all the detail in the dark sections but not blow it out in the lighter sections if you're using a constant exposure
I always do manual. In the end, you are still trying to put the entire dynamic range of the image into one combined image. Only when you rely on the camera automatic, your software has to adjust the dynamic range of each image. This means you will lose much more information then you could get with manual and RAW processing. I'd much rather try to increase the dynamic range with RAW processing. That will only work with smaller images.
Other way's to increase the dynamic range with larger images:
You could convert three sets of RAW images. One with average adjustments, one for highlights and one for shadows. You then figure out the control points and do the calculations on the average converted images. Then you replace the source files with the ones processed for highlights and run again and then do the same for the shadows. Then combine them in Photoshop by layering and painting with a soft brush with large radius to have one or the other dominate.
You could also bracket the exposure. This is possible for panoramics too. Autopano will find out the control points correctly (I think). You run autopano on the entire set. Then you remove all images but the underexposed and continue. Then you do the same for normal and over exposed and you should get 3 processed images that line up exactly.
In fact I have once used PTAssembler to stack 3 normal images I bracketed, when I didn't have a tripod. I could align all 3 images exactly and create an image with a high dynamic range without having a tripod.
Answering some questions
Peter has asked some interesting questions over here that I would like to address.
Here are some answers:
Were you only using your 50mm lens for this like you described in your tutuorial?
No, I was using my 70-200mm lens at 200mm, because I had to zoom in quite close. The result is still a wideangle (large Field of View), because there are quite a few pictures stitched together. One of the problems you will face though is the compression in depth of field a telephoto lens will cause.
When you are taking the pictures do you ever overlap so there will be 3 pictures with the same stitch joint in it?
The pictures overlap by about 20% in either direction. This is necessary, otherwise you couldn't make sure you won't have holes (the images get slightly warped to fit) and since it would be impossible to line them up exactly, plus the software needs the overlap area to figure out how the images go together.
join them in autopano it makes it rather soft
All Autopano does is figure out the control points (where the images overlap). The actual blending is done by either Enblend or Smartblend.
In PTAssembler you should turn off Feather (marked red in this part of the tutorial) plus you might want to set the Interpolator to Sinc1024.
do you usually just use the picture right out of the program (cropped, ofcourse)
I use the pictures without any adjustments, but don't crop them either. You will just crop away the information Autopano and Enblend so desperately need.
finally, do you always force it in manual exposure or do you have success with joining files that are auto metere?. just because i would think sometimes the whole scene is too constrasy to keep all the detail in the dark sections but not blow it out in the lighter sections if you're using a constant exposure
I always do manual. In the end, you are still trying to put the entire dynamic range of the image into one combined image. Only when you rely on the camera automatic, your software has to adjust the dynamic range of each image. This means you will lose much more information then you could get with manual and RAW processing. I'd much rather try to increase the dynamic range with RAW processing. That will only work with smaller images.
Other way's to increase the dynamic range with larger images:
In fact I have once used PTAssembler to stack 3 normal images I bracketed, when I didn't have a tripod. I could align all 3 images exactly and create an image with a high dynamic range without having a tripod.