 |
|
|
|
|
|
 |
| ||||||||
 |
 
|
|
|
|
|
|
|
|
|
| ||||||||
|
|
| 
| User list | You are not logged in.
When shooting panos, sometimes I fill the buffer and have to wait for it to clear. And sometimes I move a short distance and take a new pano. So the time between panos may be less than the time between shots in a pano, and the automatic building of groups by time/number of shots doesn't always work. And sometimes you just have to reshoot one position because of tourists, etc.
I could take a dark picture (leave the lens cap on, or set the time to 1/1000 and aperture to f/22) between panos, or even better, take a shot of a grey card or white balance target before shooting a pano.
If APP could somehow recognize this shot, it could automatically separate panos.
1. just looking for a dark shot wouldn't work, sometimes you over-bracket intentionally. APP would have to look for a speed greater than (some user value) or better, it should look for an aperture change (I generally shoot panos on Av or M.) That would signal the start of a new pano.
2. The user would shoot some white balance target. The user would select one sample of this target, and APP would look for it and separate panos at that target. This should be done on the center part of the image, and assume that the target will be handheld and change in angular size because it will be held at different distances, angles, etc. Or to make it simple, just put a coffee filter over the lens, take a shot, and APP could look for an all-one-color image (all gray wouldn't work, the WB may be off).
Offline
hankkarl wrote:
If APP could somehow recognize this shot, it could automatically separate panos.
Clap shots !
(No doubt the future Kolor Panocam will include a clap button)
Offline
That was suggested a long time ago, as far as I can remember and it still a great idea! Also, it may be a good addition to use the "cap-on shots" as signals for beginning and end of a panorama, as most of the times panos are mixed with single shots.
Offline
[bo] wrote:
That was suggested a long time ago, as far as I can remember and it still a great idea! Also, it may be a good addition to use the "cap-on shots" as signals for beginning and end of a panorama, as most of the times panos are mixed with single shots.
just don't do the "cap on" shot in Av mode, or you'll wait for a long time 
Offline
Yes, this notion of 'how to separate group at shooting time' is real challenge but some trick already exists :
* The standard hand over the lens idea and shoot ( you'll get a quite black image with some finger on it ). I guess you won't mislead this image with one from the panorama.
( Some people are using a piece of wood / paper / tissus, whatever ).
* Wait enough time ! If you setup the time laps / group size goal well in the 'browse folder' dialog box, you can directly separate the groups using this trick.
Imagine we use 30s. Let's assume that if we have 2 images separated with that time lapse, then cannot belong to the same panorama. So just set this value in the dialog box and put 2 in the group size goal ( so that this end rule for group creation doesn't count, only the time ). It will work.
New ideas :
* Clap shots : Yes, we could imagine to code something that detects automatically an image with 'a standard form' : it could be histogram based, form based, sift based, etc.
* I had once this idea when app creates groups based on the previous 2 rules. It often stops with some groups where 2 or more panoramas are presents. We could easily code on a right click entry something that says : divide it once more => resulting in 2 groups. That could work too and be quick.
Offline
AlexandreJ wrote:
* Wait enough time ! If you setup the time laps / group size goal well in the 'browse folder' dialog box, you can directly separate the groups using this trick.
Imagine we use 30s. Let's assume that if we have 2 images separated with that time lapse, then cannot belong to the same panorama. So just set this value in the dialog box and put 2 in the group size goal ( so that this end rule for group creation doesn't count, only the time ). It will work.
But I don't want to wait!
AlexandreJ wrote:
New ideas :
* Clap shots : Yes, we could imagine to code something that detects automatically an image with 'a standard form' : it could be histogram based, form based, sift based, etc.
* I had once this idea when app creates groups based on the previous 2 rules. It often stops with some groups where 2 or more panoramas are presents. We could easily code on a right click entry something that says : divide it once more => resulting in 2 groups. That could work too and be quick.
How about a standard form based on sift? Or better, two or three? That way we can kill two birds with one stone if one "standard form" is set to the tripod head (nadir shot):
1. signals end of pano if you shoot from the top to the bottom.
2. by shooting the nadir and cutting out the tripod in PS, APP could know what the tripod head looks like and could automatically mask out the tripod and head so that no CPs would be put on it, and it wouldn't be blended.
In this case, you need at least two forms -- one for the tripod head, one for the gray card/wb target if you do a partial pano or circular pano, and probably also one for when you forget the gray card and have to use your hand
Last edited by hankkarl (2010-03-02 17:16:12)
Offline
hankkarl wrote:
But I don't want to wait!
Then you could change the camera time shifting it towards the future...
Offline
AlexandreJ wrote:
New ideas :
* Clap shots : Yes, we could imagine to code something that detects automatically an image with 'a standard form' : it could be histogram based, form based, sift based, etc.
Tu pourrais faire des essais avec SIFT pour trouver un motif qu'il reconnait facilement même quand il n'occupe qu'une faible partie de l'image et qu'il est un peu flou. Ensuite tu le mets en ligne, on l'imprime en format carte de crédit, et on le photographie entre chaque pano (il peut même y avoir une face "début de pano" et une face "fin de pano"!)
Beaucoup de travail pour pas grand chose?
Offline