Some generic thoughs on Autopano (rant-ish)  

All things panoramic.
An open forum for every topic on photography, panoramas, IT or related.
User avatar
[bo]
Member
 
Topic author
Posts: 1226
Likes: 0 post
Liked in: 0 post
Joined: Fri May 05, 2006 8:16 am
Location: Bulgaria
Info

Some generic thoughs on Autopano (rant-ish)

by [bo] » Fri Jan 05, 2007 10:34 am

Looking at the topics from the last week or so, I see plenty of high level discussion. Users debating over the inner workings of APP and such. And I wonder - what is the motivation? This is not bug reporting for the most part and when it is, it's regarding a thing almost nobody encounters.

In my practice I try to have low RMS, acceptable geometry and nice looking colors. That takes a few minutes a project and I'm done. A colleague of mine uses APP in almost fully automatic mode (drop images, add verticals, render as psd) and is perfectly happy with the results. There are some projects that require more attention due to a human mistake while shooting, but those are exceptions and APP handles all reasonably well.

So let's get back to the question - what drives you guys to delve into algorithms, coefficients, dissect project files, etc? Maybe it's the constant search for perfection?

Which leads me to the other issue that got me wondering. I see on a regular basis projects with 30, 50, 100 or even more high MP images. The biggest pano I've had to stitch was something like 60 8MP files, of which I used 25, as there was much overlap. Most of the time I do 3-6 8MP or 13MP images and that's pretty much enough for any media purpose. Starting from a thumbnail on a web page and up to a offset printed poster.

What do you do with those huge projects you stitch?



And another thing - it is becoming more and more obvious we need a AP Lite and AP Pro separate versions. I've seen on several occasions users saying things like "you need better sift quality, as that sells" or "greater lens correction optimizer something will look nice on the box". That's nonsense. Some of the guys that work in my office and use APP daily don't even know what SIFT is and couldn't care less for optimizer algorithms or stuff like that. What the average user wants is a clean interface and nice looking panos with less hassle.

And most of the proposals here are "add this or that in the options". So Autopano actually needs stuff like to become smarter, so it can automate more tasks, a clever anti ghosting and a host of projections, all this under the hood of a cleaner, simpler interface.
Some of my panoramas, posted in the Autopano Pro flickr group.

no avatar
rkarpa01
Member
 
Posts: 21
Likes: 0 post
Liked in: 0 post
Joined: Wed Aug 30, 2006 8:14 pm
Location: Cuxhaven, Germany
Info

by rkarpa01 » Fri Jan 05, 2007 1:29 pm

Hi all,

bo, I agree with all you said (often couldn't follow the discussions). What I don't agree with is the separation of a lite and pro version. This makes things complicated and versions might drift apart. I see the need for advanced options and these might be able to include in all versions but selectable in the preferences.


Bye, Ralf.

no avatar
1dmk2
Member
 
Posts: 54
Likes: 0 post
Liked in: 0 post
Joined: Tue Jan 02, 2007 4:40 pm
Info

by 1dmk2 » Fri Jan 05, 2007 3:36 pm

iam new here, infact only bought it 2 days ago. i have no experience of any sort of pano photiography/ software. so i aggree with bo.
what i would like to see develop from these extra smart people is to share the knowldege for producing good results. ie. good work flows

simon

no avatar
GURL
Member
 
Posts: 2943
Likes: 0 post
Liked in: 0 post
Joined: Tue Dec 06, 2005 1:57 pm
Location: Grenoble
Info

by GURL » Fri Jan 05, 2007 3:40 pm

Most peoples use MS Word but will never write anything but short memos or single page letters.
Are they stupid because they never use more than 5% (at most!) of the many tools MS Word includes?

YES:

- MS Word is expensive and bulky.
- MS Word interface is cluttered with useless tools (for most peoples, that is) which make simple tasks much more difficult.

NO:
- every people (in a company, in a school, etc) sharing the same tool makes easier each person reaching the skill level he needs (there is alway somebody somewhere who can show you how to add a footnote or how to align decimal separators in a table when you need that.)
- this avoid long (thus expensive) discussions to decide who needs or don't need to add footnotes or to align decimal separators.

You could find counter-examples like Adobe Photoshop Elements against Adobe Photoshop "full version". As it's more expensive for Adobe to produce and market two different versions of the same software, the real question is: how many users of Elements bough it because they believed it was easiest to use than the full PS CS version?

Frankly, I don't know the right answer.
Georges

User avatar
[bo]
Member
 
Topic author
Posts: 1226
Likes: 0 post
Liked in: 0 post
Joined: Fri May 05, 2006 8:16 am
Location: Bulgaria
Info

by [bo] » Fri Jan 05, 2007 4:32 pm

I didn't really wanted this to be a Pro/Lite discussion... and I'm not sure whether this is a good road to follow. Maybe a "wizard"-kinda thing, I don't really know. But that's not the main point. What I see as a trend (hence the first two questions) is the will and desire of users to work more for the same result. I'd have a simple working smart application any day, instead of a super-app that does it all with 300 settings and a checkbox for every algorithm inside.

I really suggest to all interested forumers to get familiar with the Getting Real book by 37Signals. Similar principles are behind the already noted PS Elements, behind Aperture, Lightroom, MS Expression modules...

The Word example is very nice - the Word 2003 interface is practically a heap of features that has been piled up (mostly without any structure) for many years. That's why the 2007 Office has a completely new interface. My grandpa is currently typing his autobiography in Word 2007 and it took me 30 minutes to explain it to him.

People does not want a complicated tool that does a simple job, they need a simple way to do complicated things, that's a fact.
Some of my panoramas, posted in the Autopano Pro flickr group.

no avatar
Al
Member
 
Posts: 98
Likes: 0 post
Liked in: 0 post
Joined: Sat Jan 21, 2006 1:35 am
Location: Ann Arbor, MI
Info

by Al » Fri Jan 05, 2007 4:59 pm

Hi Bo,

About one of your questions, 'what do people do with such huge files?' Two uses come to mind almost immediately. The first is large prints. I recently made a 72" x 24" print of the hiking trail pano I posted a few weeks ago. The client wanted a tall photo for the space at the top of a very tall stairwell. Without interpolating pixels, that comes to ~26,000x8,600, or about 223 MP. Another application is hyper-detailed web photos that allow users to zoom through the space.

The most frequent comment I hear about prints I've made with very large data files is that people are astonished at the level of detail in them, similar to what one would get using a large format camera with 5x7 or 8x10 negatives. For me the real beauty of APP is that it allows me to do landscape photography with light, easy to carry equipment that creates image quality that is amazing.

User avatar
tvkas
Member
 
Posts: 139
Likes: 0 post
Liked in: 0 post
Joined: Thu Dec 07, 2006 6:29 am
Location: Reno, Nevada, USA
Info

by tvkas » Fri Jan 05, 2007 6:47 pm

Bo,

In the kind of publication work we do a large High DPI, large format image makes for - say - cover photograph or a full page spread. We don't use any 72 DPI of just a few MP for any printed piece.

I have zero complaints on APP. In fact when I first found it I loved it. I am in a position to learn to use APP to meet my needs rather than have APP tweaked to meet my needs. It's like asking somebody to redesign a screwdriver to work as a spanner.

BTW, the first cover shot I make will be posted right here for all to see.
Canon S5IS, 8 Megapixal
Nodal Ninja head
Lots of vodka.

no avatar
GURL
Member
 
Posts: 2943
Likes: 0 post
Liked in: 0 post
Joined: Tue Dec 06, 2005 1:57 pm
Location: Grenoble
Info

by GURL » Sat Jan 06, 2007 12:53 pm

'[bo wrote:']Which leads me to the other issue that got me wondering. I see on a regular basis projects with 30, 50, 100 or even more high MP images. The biggest pano I've had to stitch was something like 60 8MP files, of which I used 25, as there was much overlap. Most of the time I do 3-6 8MP or 13MP images and that's pretty much enough for any media purpose. Starting from a thumbnail on a web page and up to a offset printed poster.

What do you do with those huge projects you stitch?

Bernhard Otto Holtermann (sponsor) and Charles Bayliss (photographer) - 1876 - Sydney, Australia.

The masterpiece which concluded their cooperation was a negative of Sydney harbor measuring 150 x 90 cm (60" x 35") using a wet plate process. Holtermann had a 22 m (72 feet) tower built. A 3 m (10 feet) camera using a 2.50 m (2500 mm) lens was placed on the top. Bayliss and Holtermann prepared and sensitized the glass plate and took the photo while being in the camera.

During 1876 year Holtermann brought this giant negative and a Sidney panorama measuring 9 meters (30 feet) at the international Philadelphia Centennial Exhibition and other places like San Francisco and Paris (History of Photography: From 1839 to the Present - Beaumont Newhall.)

That there is no established market for such large images is unquestionable.

When, using a viewer, peoples look at my spherical panos (usually 32 million pixels and 3 megabytes), they are often somewhat disappointed when zooming-in because details are fuzzy or zooming is limited. Doubling the zooming power needs a fourfold increase in pixels. A few years ago, a 3 megabytes full-screen panorama was impossible to display. IMHO, that there is presently no gap between supply and demand for that is questionable.
Georges

User avatar
black nanar
Member
 
Posts: 130
Likes: 0 post
Liked in: 0 post
Joined: Tue Sep 05, 2006 11:15 pm
Location: on dirait le sud...
Info

by black nanar » Sat Jan 06, 2007 3:26 pm

what to add ? defish !

people ask for more features they can manage, for everything
buy a Porsche Cayenne 4WD but never get offroad
buy MSWorld just to type memos
you own a powerful system, your supposed to handle it in others mind, and that's enough for your happiness

APP is willing toward an easy and comprehensive use, let it be
remind me what will be expected date with defish ?

User avatar
[bo]
Member
 
Topic author
Posts: 1226
Likes: 0 post
Liked in: 0 post
Joined: Fri May 05, 2006 8:16 am
Location: Bulgaria
Info

by [bo] » Mon Jan 08, 2007 10:15 am

tvkas, yes, I know that - my main business is advertising and print production.

GURL, thanks for the info :)

I know there are certain applications of such gigapixel images, just did not expect that there will be so many users pushing the limits. When I (struggled to) used Hugin, the community was mainly 360x180 guys, on-screen purposes, etc. Even for print - for a 50x70 standard poster at 200dpi = 70MB file is quite enough. We did an exposition in our embassy in Vienna with 100x70 posters and we used 150MB files. This means a set of 15 to 20 8MP images should be enough for pretty much everything in that direction.

What's why such 400-photos panoramas (or even more) amaze me ;)
Some of my panoramas, posted in the Autopano Pro flickr group.

no avatar
GURL
Member
 
Posts: 2943
Likes: 0 post
Liked in: 0 post
Joined: Tue Dec 06, 2005 1:57 pm
Location: Grenoble
Info

by GURL » Mon Jan 08, 2007 1:10 pm

'[bo wrote:'][..] just did not expect that there will be so many users pushing the limits.
[..] What's why such 400-photos panoramas (or even more) amaze me ;)

In their first attempts, too many users - me included - tend to use a huge number of images. A two images pano is much faster to learn how to use a stitcher.

I'm often afraid about a new APP user browsing the forums and concluding that shooting a very large number of images is the way to go. This is the reason my first post was about the future APP light version. Perhaps, having a "new users" (or you name it) sub-forum would help?
Georges

User avatar
[bo]
Member
 
Topic author
Posts: 1226
Likes: 0 post
Liked in: 0 post
Joined: Fri May 05, 2006 8:16 am
Location: Bulgaria
Info

by [bo] » Mon Jan 08, 2007 3:04 pm

GURL wrote:A two images pano is much faster to learn how to use a stitcher.

Yes, my thoughts exactly.

I don't want to dictate people how to use the software; nor I want to split them into "new" and "old" users. It just startles me that many of them are making high-end complicated demands for very specific edge cases, while the general use seems to be left a bit on the side... For example, if I (we) haven't pushed that much a while back for a anti-ghosting blender, we could've still had Multiband only. In the same time Panorama Fanatic was dissecting project files and reporting issues that sounded like the end of the world, yet there were no other complains on those counts.

CP Editor was nice addition, but instead of adding a bunch of extra manual features to it, maybe improving the logic to do its work without the need of user intervention would be a lot better. I mean, I don't really want to mess with CPs. And I don't - on 80% of the projects I get acceptable results without opening the CP Editor. My thinking is: "let's try to reduce those 20% that need manual intervention via smart application logic"; as opposed to the much requested "give me more tools and options, so I can work on those 20% and the rest 80% myself".

"It just works" is way better than "I just have to make it work".

ps: What I really would like to see, for statistical purposes (and for the sake of the discussion) is a silent reporting module for APP what builds a DB online. So we can see what people are really using the software for - number of files, output format and final size, blender, type of projection, fov, hdr or ldr, etc.
Some of my panoramas, posted in the Autopano Pro flickr group.

User avatar
AlexandreJ
Kolor Team
 
Posts: 5987
Likes: 7 posts
Liked in: 10 posts
Joined: Mon Nov 14, 2005 4:56 pm
Location: Francin, France
Info

by AlexandreJ » Mon Jan 08, 2007 3:55 pm

'[bo wrote:']ps: What I really would like to see, for statistical purposes (and for the sake of the discussion) is a silent reporting module for APP what builds a DB online. So we can see what people are really using the software for - number of files, output format and final size, blender, type of projection, fov, hdr or ldr, etc.

I'd love to see that, but that's a nightmare in customer relation. Everytime a software developer want some statistical feedback, it has to get through customer privacy defense league. People don't wont software to spy their use even for design or workflow improvement.

User avatar
AlexandreJ
Kolor Team
 
Posts: 5987
Likes: 7 posts
Liked in: 10 posts
Joined: Mon Nov 14, 2005 4:56 pm
Location: Francin, France
Info

by AlexandreJ » Mon Jan 08, 2007 4:13 pm

I wonder if we should not setup some pool to get some figure on some questioning here ...
What should be the next big thing to improve or add in autopano :

- More automatic in more case
or
- Give user more toys to fine tune the panorama

User avatar
tvkas
Member
 
Posts: 139
Likes: 0 post
Liked in: 0 post
Joined: Thu Dec 07, 2006 6:29 am
Location: Reno, Nevada, USA
Info

by tvkas » Mon Jan 08, 2007 5:03 pm

You could put the report mod in but give users the ability to turn it off. It skews the results a little but would still give you some ability to see how it is being used.

As for any improvements I can't think of any except to say that the users here seem to be very experienced with APP. Allowing users the ability to use more manual features I am sure would make the experienced users happy. Perhaps making the GUI have 2 settings, Normal (like we see now) and experienced (which allows people to have extra settings).

I do agree with black nanar - buy MSWord to only type memos. But some will be able to type some very fancy memo's. APP works so well right now. Don't fix it too much.
Canon S5IS, 8 Megapixal
Nodal Ninja head
Lots of vodka.

no avatar
GURL
Member
 
Posts: 2943
Likes: 0 post
Liked in: 0 post
Joined: Tue Dec 06, 2005 1:57 pm
Location: Grenoble
Info

by GURL » Mon Jan 08, 2007 5:34 pm

AlexandreJ wrote:- More automatic in more case
or
- Give user more toys to fine tune the panorama

I'm afraid it's not that simple.

Provided the results are identical, any sound photographer should prefer an easy and fast tool to a more difficult to use and slow one. Furthermore, an easy and fast tool lends to more panos being actually shot and stitched and we all need to improve our skill and art this way (few of us, if any, have a long practice of panoramic photography.-)

The problems arise when the result we get don't come up to the high expectations we had when shooting. Whether this is caused by RTFM, improper shooting or the subject being actually difficult, this happens to any of us.

My suggestion would be to ask APP users to send you the source images of the panos they were heavily disappointed not being able to stitch, whichever the reason (APP bugs being excluded, that's another matter.)
Georges

User avatar
[bo]
Member
 
Topic author
Posts: 1226
Likes: 0 post
Liked in: 0 post
Joined: Fri May 05, 2006 8:16 am
Location: Bulgaria
Info

by [bo] » Tue Jan 09, 2007 8:40 am

AlexandreJ wrote:
'[bo wrote:']a silent reporting module for APP what builds a DB online

Every time a software developer want some statistical feedback, it has to get through customer privacy defense league.

Yeah, I know that. Maybe if it is disabled by default and you have to enable it, thus agreeing to report anonymously this and that to a central server. Like Winamp asks you while installing.

I don't think a poll will be sufficient - right now there are more high end users writing here and the exposure to users will not be optimal. I know of at least three users that never posted here but are doing quite a bit of panoramas - they just don't have any problems.

Georges' suggestion seems nice, also. A database of cases, in which APP does a poor job, each case illustrated by an image set.

The main idea behind all this is to know how actually the software is used and what are the problematic areas (not bugs) in panorama composition.
Some of my panoramas, posted in the Autopano Pro flickr group.

User avatar
AlexandreJ
Kolor Team
 
Posts: 5987
Likes: 7 posts
Liked in: 10 posts
Joined: Mon Nov 14, 2005 4:56 pm
Location: Francin, France
Info

by AlexandreJ » Tue Jan 09, 2007 9:10 am

'[bo wrote:']a silent reporting module for APP what builds a DB online

'[bo wrote:']The main idea behind all this is to know how actually the software is used and what are the problematic areas (not bugs) in panorama composition.

Hum, yes. Activation would be required. The main discussion is now what should be collected :
- stats on usage for each button / tools
- stats for rendering option
- stats on panorama type ( number picture / input file format / camera used / lens used )
- As the goal is to find out how people are using the software in this way, having like a timeline record of tools used from detection to rendering would be nice too
- what else could be useful ?

'[bo wrote:']I don't think a poll will be sufficient - right now there are more high end users writing here and the exposure to users will not be optimal. I know of at least three users that never posted here but are doing quite a bit of panoramas - they just don't have any problems.

I agree that in the community here a great part is composed of high end users :) It's not representative of customers, globally.

GURL wrote:A database of cases, in which APP does a poor job, each case illustrated by an image set.

That's is the aim behind the ftp server : having a way to transfer the case to us. I've sorted them here and now it's a pretty big database ( around 50 gigas of input files ).
I made some classification : issue on input format, issue on stitching, rendering, etc. So when releasing a new version, we have a quick way to check if no regression occured and we can check if issue are solved. This method really improved the software globally. Of course if any user have some new test cases to share, please, just do it. We are really enjoying them to see where the problem are lying, what is missing in APP, how we could improve the workflow. So a definitive YES : this database exists, just continue to provide any test case to improve it.

no avatar
beeloba
Member
 
Posts: 1930
Likes: 0 post
Liked in: 0 post
Joined: Fri Aug 18, 2006 10:51 pm
Info

by beeloba » Tue Jan 09, 2007 11:39 am

'[bo wrote:']What do you do with those huge projects you stitch?

It as been quite a long time since big inkjet printers with RIP (raster image processors) were making huge prints with no need of giga pixels files.

The interpolation issue is a basic user problem, visual resolution can be set at 1200 to 4000 dpi and frames can't be seen unless you look at a 20 square meter image at less than 50 centimeters, which is not the usual purpose.

Here are some enlargements made. This is not the biggest I've seen, just examples.

Do you think those pictures have been made with 20 shots:
http://www.imagingpath.com/prod_images/NJ1000i%20Printer_CMYK.jpg
http://www.bespoke.co.uk/clients/Vutek/images/SouthernCrossStation300.jpg
Last edited by beeloba on Wed Jan 10, 2007 4:03 pm, edited 1 time in total.

User avatar
[bo]
Member
 
Topic author
Posts: 1226
Likes: 0 post
Liked in: 0 post
Joined: Fri May 05, 2006 8:16 am
Location: Bulgaria
Info

by [bo] » Tue Jan 09, 2007 11:53 am

Heh, we're printing 5x3 meter vynil billboards on a regular basis. Most printers require a 300dpi file in 1:10 scale. This means 50x30 cm at 300dpi. This is more than enough, as the closest you can get to a billboard is 4 meters. Some new printers that use alternative media and solvent inks use bigger files - 1:1 scale @ 100 dpi, which means a 600MB+ TIFF file. But somehow you fail to persuade me the majority of users writing here are using APP for such purposes :) !
Some of my panoramas, posted in the Autopano Pro flickr group.

no avatar
steve
Member
 
Posts: 48
Likes: 0 post
Liked in: 0 post
Joined: Tue Apr 25, 2006 8:51 am
Info

by steve » Tue Jan 09, 2007 4:34 pm

Hi [bo]... I can give you a quick and common reason why I'm one of the people making great demands of Autopano. I'm trying to start a company doing panoramic photography, and I need to be able to ensure results for (almost) each and every one of them. I was completely unable to make Autopano handle and process every 360 degree panorama. It would frequently split the panorama into partials or have a hole where it failed to match against white walls, or a big burning gap covering half the horizon where the sun is.

When I have these problems, the only thing I can do is to manually fix them. Autopano is (still) very poor for this, as it calculates minimum changes which results in control points not moving something from _here_ to _there_... which is my most frequent activity. My current workflow consists of Hugin for placing control points, and an old version of Autopano (1.2.1) to import PTO files... because that's the only version which lets me turn off any attempts by Autopano to calculate.

So my requests for more manual control are because I don't do this as a hobby. I have plenty of wonderful panoramas that came out very well, but I also have plenty of panoramas that were not good enough to sell. Delivering to the customer means that even if the pictures are not the best, they still need to be complete functional 360 degree panoramas without huge holes, burning gaps, insane wobbles and bizarre distortions from stretching across a flock of flying birds.

So the requests are a matter of trying to ensure reliability of results.
Last edited by steve on Tue Jan 09, 2007 4:40 pm, edited 1 time in total.

User avatar
[bo]
Member
 
Topic author
Posts: 1226
Likes: 0 post
Liked in: 0 post
Joined: Fri May 05, 2006 8:16 am
Location: Bulgaria
Info

by [bo] » Wed Jan 10, 2007 9:50 am

I see your point. But I'm pretty sure you would prefer an application that does what you need by itself, instead of forcing you to do the work. I do panoramas as a part of my business too and my thinking is that I should spend *less* time fiddling with APP or any software for that matter. Drop some images and let it render, instead of spend the day straightening verticals and trying projections.

Manual controls a somewhat better as a short term solution, as they give advanced users the ability to fix what the software cannot. However:
1. Casual users rarely use those tools
2. Such solutions create much legacy problems - I don't imagine any manual tool being removed after much work was put in it, even if better automatic counterpart exists
3. Those tools require time to code and debug, then support. Time better spent on making the application smarter.

That was the main reason I discussed modular (or plugin) architecture for APP almost an year ago. You may need manual CP Editor - fine, but I don't and as it will not be a part of the application core (as well as most advanced manual stuff), the focus will fall naturally on the "AUTO" part of Autopano. Lest we rename it to Manualpano?

Just dreaming here, excuse the rant :)
Some of my panoramas, posted in the Autopano Pro flickr group.

no avatar
steve
Member
 
Posts: 48
Likes: 0 post
Liked in: 0 post
Joined: Tue Apr 25, 2006 8:51 am
Info

by steve » Wed Jan 10, 2007 2:33 pm

Sure, it would be nice if I could just drop pictures in and it would work... but that just isn't going to happen. The basic information is missing from the photographs and the human brain is using many other methods to work out what should have been there. Autopano is using a very good, but single, method with the many adjustments by AlexandreJ almost being a neural net selection.

If you want to see a totally automatic panorama tool, have a look at Calico. That's a drag'n'pray tool for those who can say "oh well, that one sucks, let's try something else". Autopano can aspire to that, or it can handle reality and provide the tools to correct for situations it cannot handle. I'm OK with a program that says "I'm a mere computer, I need some help here".

Where Calico and some of the aspects of the current Autopano are a problem is when it takes your input, digests it, and then delivers the result on your new carpet with a goofy grin and expects a pat on the head. :)

The Dev Plan is exactly right the way it is. The stuff to support features is being added, then the majority of my problems will go away with 1.4 and fisheye support. I use Hugin for control points because I'm generating 85x135 photos from DeFish, which is outside of what Autopano can handle.
Last edited by steve on Wed Jan 10, 2007 2:38 pm, edited 1 time in total.

User avatar
AlexandreJ
Kolor Team
 
Posts: 5987
Likes: 7 posts
Liked in: 10 posts
Joined: Mon Nov 14, 2005 4:56 pm
Location: Francin, France
Info

by AlexandreJ » Wed Jan 10, 2007 3:14 pm

To Steve. I just don't see why you should use hugin to fix something. We have a far better tool with our control point editor. I'm really curious to see some of your panorama as the only case we cannot handle yet is the full blue picture in the sky (but it's planned for next version).

Our design and goal for autopano is still having the most automatic stitcher with the possibility to have manual control.

steve wrote:... the current Autopano are a problem is when it takes your input, digests it, and then delivers the result ...

Please develop ! We can surely improve stuff if you tell us where we should have more work.

no avatar
GURL
Member
 
Posts: 2943
Likes: 0 post
Liked in: 0 post
Joined: Tue Dec 06, 2005 1:57 pm
Location: Grenoble
Info

by GURL » Wed Jan 10, 2007 3:38 pm

steve wrote:Sure [...] ... but that just isn't going to happen. The basic information is missing from the photographs and the human brain is using many other methods to work out what should have been there.

APP (and other stitching tools) are prensently accomplishing some tasks which looked impossible some years ago. Computer programs are not "intelligent", but brute force, ad-hoc solutions and subtle strategies sometimes result in human intelligence specificity having to be updated (chess game, after half a century of hard work, is an interesting example.)

I sometimes wonder whether absence of manual CPs placement is an incentive to improve SIFT and/or finding some other new solutions. In some situations, being able to modify X, Y values of a single point in the link editor displayed table would help :rolleyes:

In the following example APP ignored the leftmost source image. Using the link editor I forced APP to add this image, selected a small area where it found some good and some misplaced CPS. I removed the misplaced ones. Smartblend (BTW, who dreamed of Smartblend before Enblend suprised us with its incredibly nice seams?) did the rest. Now I'm using the layer editor and PS to remove clouds where I want some mounts being more visible...

:) : I removed duplicated CPs from the list but not having to do that would be nice.




Georges

Next

Who is online

Users browsing this forum: No registered users and 2 guests