APP 1.4 try to find links between all images. So you have a lot of links with bracketed panos.

With APP 1.9 (future 2.0), you can ask him to do the same, or to only find links between images in one exposure (don't know which exposure it uses) + links in each stack (between bracketed shots).

+1  +-+-+-+-+-+-+-+
    | | | | | | | |
 0  +-+-+-+-+-+-+-+
    | | | | | | | |
-1  +-+-+-+-+-+-+-+

or

+1  + + + + + + + +
    | | | | | | | |
 0  +-+-+-+-+-+-+-+
    | | | | | | | |
-1  + + + + + + + +

I haven't tried this feature yet...

I tried many times with APG-1.9-a1 and a3, and I tried again yesterday but I had no luck getting stacks to work. However, proper stack support for expanding dynamic range and for focus stacking is planned for APP-2.0.

I just hope that stacking will also work in the absence of EXIF data... when I do macropanography there is no EXIF data.

I looked at the documentation, at http://www.autopano.net/wiki/action/view/Latest_Beta but that uses 0.0 as its example.

Let's ask Alexandre to be more specific about stacking in documentation with the next release. What should be covered?

Certainly: if there is EXIF data what field in that data is used to construct stacks.

Also, indeed, what happens if there isn't any EXIF data, and how would one build stacks manually.

As an extension to this, if settings have to be tweaked manually - in groups, are the fields displayed, in the Settings feature of APG, which would need to be tweaked; or will EXIF data have to be preprocessed, i.e. using another application?

btw: when I got stacking to work, it was against images with EXIF data shot using CANON AEB with RAW (.cr2 images) converted to jpg using DPP. This, of course, produces 3 stacks with bracketing symmetrical around a mid-point, in my case 0.0.

Hi, sorry about this but I used the wrong area to post a reply with an image attached.

Hi fma38 and everyone else,
Am I right in assuming that in APP ver 2.0 the stack facility is or will be as per the enclosed diagram. That is, a bracketed group of exposures (-1,0,+1) with focus at 6 metres, and a second bracketed group focused at 16 metres will be treated by the algorithm as one stack containing good dynamic range and extended depth of field, AND all this will be performed automatically. The algorithm will then go on to treat stack 2 equally and the stitch both stacks together into a pano.
It is just that I am trying to get my head around the concept so that I can take steps to develop a technique that will work.

Okay, I think the enhanced focus depth is more given some confusion to the first aim of this thread. So let's me explain first how stacks are working.
In fact, it's like fma38 said in this post : http://www.autopano.net/forum/p30156-20 … -40#p30156 :
- either all level of the bracket will be linked by cp,
- or just one level.
Inside a bracketed serie, the links will always exists. But this is only tells how the CP detection will work with stacks.

Now what can we do with a stack : in group the view, when a stack is detected, you will be able to use "N to 1" plugin. This kind of plugin will use N images in input and outputs one image. There are several such plugins planned ( not all for the final december release, ok ). For example : tufuse or enfuse could be such a plugin. We have our own fusion technology that will be able as "N to 1" plugin. Normally, 2 plugins will be present for 1st release : Fusion, Merge to HDR to create a real 32bits HDR ( perhaps we'll have the time for a command line plugin to call external tools like enfuse from APG directly ).

But you'll have also another workflow : stitch then fuse by using the layer tool. The "N to 1" plugin could be called after the stitching too. So both possibilities : hdr/fuse before the stitching or hdr/fuse after the stitch.

Now, focus enhancing : Max Lyons did it in a second pass after or before the fusion, but that's just another pass. It is easy to do in the fusion tool too. That's in fact how we did it internally. But our aim was first to put it in the multiband blender, so that even panorama without any stack could benefit of that.

A point of detail that may help, if you do not know of it. There is a setting 'force every image to be in the same panorama', which applies to detections run after images have been split up into groups.

This is available, in global settings from the main window at Edit | Settings| Detection or under the Settings icon within a group (the spanner and screwdriver one) see http://www.autopano.net/wiki/action/vie … Detection.

btw and generally, I found the link in the wiki to the detailed documentation of settings quite hard to find - and I am pretty familiar with the documentation!

Last edited by leedsjoe (2008-10-06 08:29:31)