 |
|
|
|
|
|
 |
| ||||||||
 |
 
|
|
|
|
|
|
|
|
|
| ||||||||
|
|
| 
| User list | Rules | You are not logged in.
I am new to Autopano Pro. In fact i only have the trial version, and it works well so far. I'll probably buy it this week, as soon as I figure out a way through a few obstacles.
My setup is rather ridiculous, admittedly. Hey, its experimental, but works amazingly. I have an old 4X5 enlarger. I then project images down to the baseboard where I have a camera pointing up towards the enlarger lens. The camera lens is removed so that the enlarger projects directly onto the camera sensor. Now, the camera is mounted to an XY table controlled by stepper motors. The camera moves, takes picture, moves take picture etc. It is very accurate down to about 1/10mm. My point is that the grid is very well defined, and if I could figure out a way to tell the program where each image is, I am sure I would be able to help the detection a lot! I have the info, I just need to let the program know. Mind you, the automatic detection settings work really really well on a lot of material. But I run into trouble on some images I have which are mostly negative space.
From what I can tell, I cannot use the "Use Grid Position" or "Assume Row/Column shooting" because I am not using the gigapan head. That is a shame, because I am definitely using a grid, and it is row/column shooting, and its pretty damn accurate.
Anyway, the next best bet appears to be under "Optimization Stages/Advanced Settings." There I see "Local Approach." The manual says, "Lets you pre-position the images locally." I cannot figure out how it is that I can pre-position the images. Where does it allow me to do that? Sorry if that is a stupid question.
I am sure I will get a lot of questions as to why I don't just scan negatives instead of projecting onto the camera sensor. Well, I am not really using the enlarger for negatives. I actually make small drawings on glass. Sometimes intentionally scratch or crack the glass, and I get a lot of interesting results. I've tried scanning, and its just not the same. This is obviously not for commercial photography.
If someone can tell me how to pre-position the images, that would be a great help. I am used to PTGUI, which allows you to set the rows and columns. Unfortunately PTGUI is not so good with flat stitching which is why I am trying you guys. Thanks!
Offline
Welocme to the forum...
Autopano Giga allows you to Save as a template from a good stitch.
This may help - perhaps you could try it using a demo version.
It sounds as if you tried using the Gigapan Import filter? Did it not help at all?
You might also try using the free Papywizard software in simulation mode to create a Mosaic XML format data file that can be used with the Papywizard Import filter.
How many images are there in each image set? If relatively few you could just force all images in to the pano then manually position those for which no control points were acutomatically.
Last edited by mediavets (2013-01-21 10:12:47)
Online
tomasulo wrote:
My setup is rather ridiculous, admittedly. Hey, its experimental, but works amazingly. I have an old 4X5 enlarger. I then project images down to the baseboard where I have a camera pointing up towards the enlarger lens. The camera lens is removed so that the enlarger projects directly onto the camera sensor. Now, the camera is mounted to an XY table controlled by stepper motors. The camera moves, takes picture, moves take picture etc. It is very accurate down to about 1/10mm. My point is that the grid is very well defined, and if I could figure out a way to tell the program where each image is, I am sure I would be able to help the detection a lot! I have the info, I just need to let the program know. Mind you, the automatic detection settings work really really well on a lot of material. But I run into trouble on some images I have which are mostly negative space.
Hi!
Could you tell what´s the reason to do it this way? What do you want to achieve?
tomasulo wrote:
From what I can tell, I cannot use the "Use Grid Position" or "Assume Row/Column shooting" because I am not using the gigapan head. That is a shame, because I am definitely using a grid, and it is row/column shooting, and its pretty damn accurate.
Anyway, the next best bet appears to be under "Optimization Stages/Advanced Settings." There I see "Local Approach." The manual says, "Lets you pre-position the images locally." I cannot figure out how it is that I can pre-position the images. Where does it allow me to do that? Sorry if that is a stupid question.
I am sure I will get a lot of questions as to why I don't just scan negatives instead of projecting onto the camera sensor. Well, I am not really using the enlarger for negatives. I actually make small drawings on glass. Sometimes intentionally scratch or crack the glass, and I get a lot of interesting results. I've tried scanning, and its just not the same. This is obviously not for commercial photography.
If someone can tell me how to pre-position the images, that would be a great help. I am used to PTGUI, which allows you to set the rows and columns. Unfortunately PTGUI is not so good with flat stitching which is why I am trying you guys. Thanks!
You are moving the camera parallel to the negative/positive in your enlarger.
A stitcher assumes images taken from one point: the NPP (NoParallaxPoint) the camera rotates around taking pictures sequentially.
There are ways to use parallel or linear stitching very well.
Basically it´s important to know for which purpose you use that method to give advises.
best, Klaus
Last edited by klausesser (2013-01-21 14:13:29)
Offline
Thanks for such quick replies. I'm confident that you guys have the answer. First let me elaborate on what I am doing, then I will answer your questions, and hopefully we can narrow it down.
So, first I am not a commercial photographer. I'm mainly a painter, but my day job involves electronics and automation. So many times when I encounter a problem I fall back on that. So I have been working with images whereby I take a piece of laminated safety glass, and then crack it with a hammer. Or somethings I just scratch it, or hit it with a flame until it cracks. The images that result from this simple process are quite beautiful in my opinion. The way the glass cracks, the lines that result are so painterly. In some cases they look like calligraphy. So I am simply creating a series of large scale photographs based on this. So the question is why don't I just scan with a slide scanner? Well, I tried, and the images just don't come out that great. There is something about the projecting of light through an enlarger which just makes them look better. You'll have to trust me on this. With a scanner, the cracks look just like a crack in glass; wherein with the projection, the lines are dark and look more like strokes from a pen. Also there is the issue of size. The glass I use is basically car windshield glass. It is generally 6mm thick, and too thick for many slide scanners. Also, sometimes the glass can be as large as 30cm square. Thats a big negative! My XY table can scan up to 13cm in either direction. So in this case, I just move it around in the enlarger a few times, so as to get the whole thing. I've attached a few pictures of my rig, and also a test image which I put together with autopano pro. This image was taken with about a 12cm square piece of glass. Projected onto the camera sensor, it took 130 pictures to put together. Its all automated, so the all 130 pics only took about 2 minutes for all of them. Then, I stuck in autopano pro, clicked a few settings like orthographic projection, and turn off lens correction, and it came out just like this on the first try! Really great software. Now I have other images which have a ton more negative space to deal with, and the auto detect just doesn't work that great.
In answer to some questions:
>>
It sounds as if you tried using the Gigapan Import filter? Did it not help at all?
<<
Didn't know I can do that with using a Gigapan head. Is it possible?
>>
You might also try using the free Papywizard software in simulation mode to create a Mosaic XML format data file that can be used with the Papywizard Import filter.
<<
Interesting. I have no idea how to do this, and will see if I can figure it out. Thanks for the tip
>>
How many images are there in each image set? If relatively few you could just force all images in to the pano then manually position those for which no control points were acutomatically.
<<
There generally about 80-130. This may in the end be the simplest option. So I can manually reposition, and then reoptimize, and it will try to generate new control points and adjust the image? Usually, the software gets most of it right, so I can move around 20 or so wrongly placed images.
I'm also still confused over "Optimization Stages/Advanced Settings." There I see "Local Approach." The manual says, "Lets you pre-position the images locally." I still don't see how to pre-position the images in advance.
Thank you for reading through this very long post. I'm really new with this software, so pardon any really obvious questions by me. Thanks!
Offline
tomasulo wrote:
>>
It sounds as if you tried using the Gigapan Import filter? Did it not help at all?
<<
Didn't know I can do that with using a Gigapan head. Is it possible?
Yes, it's worth a try.
The Gigapan robotic heads do not record shooting positions/co-ordinates, but the Gigapan/Import filter allows you to specify the order in which the images are shot which assists the stitcher with placement.
You need to shoot in one of the supported shooting patterns.
The Gigapan Import filter assumes that the camera is mounted in landscape (rather than portrait) orientation.
Last edited by mediavets (2013-01-21 19:25:18)
Online
tomasulo wrote:
You might also try using the free Papywizard software in simulation mode to create a Mosaic XML format data file that can be used with the Papywizard Import filter.
<<
Interesting. I have no idea how to do this, and will see if I can figure it out. Thanks for the tip
The Papywizard software was developed to control a Merlin astrotelescope mount so that it can be used as a robotic pano head. (Kolor markets a packaged version under the Panogear brand).
The Papywizard software ( http://www.papywizard.org/ ) can be used in simulation mode without being connected to a Merlin mount.
Papywizard has two shooting modes, Mosiac which shoots a regular grid (which is what you should use), and Preset mode designed spefically for shooting spherical panos, which shoots are arbitrary pattern of co-ordinates defined by the user in an XML format custom preset file.
You can download the Papywizard software here:
http://www.papywizard.org/wiki/Download … ntversions
Choose the latest development version, there are versions for Windows, Mac, Unix and the Nokia N800 and N810 Internet Tablets.
Install and Configure to match the details of your camera, and experiment with settings for a lens (even though you are not using one), and camera orienation, then in Mosaic mode specify the number of rows and columns to match your shoot.
Papywizard can be set to write an XML format data file whcih records the shooting co-ordinates which is then used with the Papywizard Import filter in APG, this ensures that no images are left out even if APG cannot find control points.
I've no idea whether it will help with your unusual setup, but again I feel it's worth a try.
Last edited by mediavets (2013-01-21 19:46:50)
Online
Post #5 appears to be spot on. Unfortunately, my XY table scans the exact wrong direction for this to work. I tried all the different options. I actually need rows upward, but with #1 in the bottom right not the bottom left. I was thinking that rows downward would work, but for that I need to flip all of my images first. This is a case of making things way to complicated. Papywizard looks cool, but that my be a bit of work as well. I'll have to give a try.
Anyway, being that this issue of not being able to resolve an image is actually a rarity, rather than the norm, and the autodetect works really well, I'm just going to table the few problem issues I have at the moment, and get some work done. I'll be back, and thank you for your kind help so far.
EDIT/UPDATE:
Actually, it wasn't that hard. I just batch 180 flipped all the images. Then re-imported them into autopano, using rows downward, and voila! works great. I only have to do this once in a while, so its not a big deal... at least now I know the trick! Actually, the real trick I think is to just mount my camera the other way on the table, then all my imports should work correctly. Thanks again!!
Last edited by tomasulo (2013-01-21 21:41:30)
Offline
Just upgraded from trial to real version!
Offline
I'm glad to hear that this is working well for you. With our scanner, we move the negative and keep the camera fixed. Currently, we're working at automating the negative stage with stepper motors and an Arduino. Here's an early manual version: 
Currently, we're using a D600 or a D800 and a 55 Nikkor macro at 1:1, although there are a lot of good lens options.
Last edited by pdesmidt (2013-01-24 01:49:58)
Offline
What a great read!
Love it when people do the unconventional . Best of luck tomasulu 
Offline