gkaefer wrote:

are the images coming from a eos 5D Mark iii? if yes this case seems to be known - I saw another thread about this.
what is the goal of your pano? if you wanna create a tour for the web, then its better to convert your raws to tiffs before importing to autopano. theres absolute no loss of quality. But autopano does convert the raw images internally to tiffs and the used algorithm never can be so efficient than the rawconverter of your comera brand...

I'm using a Canon Powershot s100.
For some reason, when I import the CR2 files directly into Autopano,
the extremeties of the image contain more of the panorama than when I convert them to TIFF first.  It's very strange.  The CR2 panorama literally contains more of the landscape, it's a wider image. The TIFF version is literally smaller.  I'll upload a screenshot after work to ilustrate.
Anyway, this is why I want to be able to do the whole thing in Autopano - and I have no idea where the colour information is going.  I can't find any settings to fix the problem.

mediavets wrote:

slaughter wrote:

I'm using a Canon Powershot s100.
For some reason, when I import the CR2 files directly into Autopano,
the extremeties of the image contain more of the panorama than when I convert them to TIFF first.  It's very strange.  The CR2 panorama literally contains more of the landscape, it's a wider image. The TIFF version is literally smaller.  I'll upload a screenshot after work to ilustrate.
Anyway, this is why I want to be able to do the whole thing in Autopano - and I have no idea where the colour information is going.  I can't find any settings to fix the problem.

Apparently it's an issue with dcraw:
http://code.google.com/p/photivo/issues/detail?id=71

"the culprit is not Photivo but dcraw. Photivo has dcraw 1.445 built-in to decode raw files. Standalone dcraw 1.445 exhibits the same problem. The good news: The most recent standalone dcraw 1.447 works fine."

APP/APG RAW conversion also uses dcraw.

ok - so basically i have to wait until APG is updated with the latest version of dcraw..?

[bo] wrote:

Basically you should stick to the generally accepted process of de-rawing the RAWs in a dedicated application, IMNSHO

um, i already explained why i want to use APG to do this - i'll paste that information here again:
"
the extremeties of the image contain more of the panorama than when I convert them to TIFF first.  It's very strange.  The CR2 panorama literally contains more of the landscape, it's a wider image. The TIFF version is literally smaller.  I'll upload a screenshot after work to ilustrate. Anyway, this is why I want to be able to do the whole thing in Autopano"

Hopefully there will be an update soon which addresses the issue

mediavets wrote:

slaughter wrote:

[bo] wrote:

Basically you should stick to the generally accepted process of de-rawing the RAWs in a dedicated application, IMNSHO

um, i already explained why i want to use APG to do this - i'll paste that information here again:
"
the extremeties of the image contain more of the panorama than when I convert them to TIFF first.  It's very strange.  The CR2 panorama literally contains more of the landscape, it's a wider image. The TIFF version is literally smaller.  I'll upload a screenshot after work to ilustrate. Anyway, this is why I want to be able to do the whole thing in Autopano"

Hopefully there will be an update soon which addresses the issue

The message I referenced before suggests that the resizing is produced by dcraw so your TIFF is probably the correct version of the image?

http://code.google.com/p/photivo/issues/detail?id=71

Nope, check it out - the purple image is the CR2 panorama directly imported into APG.  As you can see, there is more of the image in this version, at the left extremity. The colour pic is from the same panorama, but after i pre-converted to TIFF first, and then imported those TIFFs into APG.  I added the black fringing, but the point is about the amount of information that is missing when I convert to TIFF first.  How is this even possible? The source JPG / CR2 files of the left extremity of this panorama does NOT have as much of the landscape as you see in the purple post-APG image.  It's like it's hidden until you import the CR2 into APG, and only APG can see it.  I'm quite perplexed.

mediavets wrote:

slaughter wrote:

Nope, check it out - the purple image is the CR2 panorama directly imported into APG.  As you can see, there is more of the image in this version, at the left extremity. The colour pic is from the same panorama, but after i pre-converted to TIFF first, and then imported those TIFFs into APG.  I added the black fringing, but the point is about the amount of information that is missing when I convert to TIFF first.  How is this even possible? The source JPG / CR2 files of the left extremity of this panorama does NOT have as much of the landscape as you see in the purple post-APG image.  It's like it's hidden until you import the CR2 into APG, and only APG can see it.  I'm quite perplexed.

What software did you use for this conversion?

I used PS CS5. As far as I know, a CR2 shouldn't contain any more of the actual image in terms of its dimensions, than its respective JPG or TIFF version.  How is it that when APG converts a series of CR2s into a panorama, it is able to detect more of the landscape than if it works with pre-converted TIFFs?
It's like the photo scanner in Blade Runner that magically detects additional landscape data which isn't otherwise present on the hard copy...

mediavets wrote:

slaughter wrote:

I used PS CS5...

With Camera RAW 6.6?

http://www.adobe.com/support/downloads/ … ftpID=5312

Yes. How on earth is it possible for my photographs to have more than a few % more landscape visible within them when imported directly into APG to create a panorama from?  no matter what I use to pre-convert to TIFF, the extra bit of landscape I see in the completed panorama is quite simply NOT there in the source files.  I'm really bummed that APG only produces a purple image when I create a panorama from my CR2 files, because that extra bit of landscape makes a lot of difference.

Here's another one.  Take a look at the row of three white boxes in the landscape, at bottom left of the image.  In ALL of the source files, CR2 as well as the TIFF conversions, there is NO source image which shows any part of the street beyond those boxes, to their left.  Yet in the final panorama - from the CR2 version - there is a whole extra section of street clearly visible!  The entire panorama is more expansive, showing landscape which isn't even visible in the source files!  What the hell am I missing here...?  and how to get to create a full colour image if APG is only able to create a purple product...??

A final look at this:  i have also attached how APG sees the JPG or TIFF of a single image, vs how APG sees the CR2 version of the same image.  The darned thing is BIGGER. WHY? When I view the CR2 in Canon DPP, OR in PS, the image is as per the colour one.  Yet when I import the CR2 into APG, APG magically sees about 10% more photograph than what ANY other software will show me.  Help!  The nightmare of this is APG saying "check it out, I will create a panorama containing, magically, more of the landscape you thought was there!  But I'm only going to let you see it in PURPLE!"

Slaughter,
IMHO, the explanation is as follows:
http://usa.canon.com/cusa/support/consu … ifications

The S100 has a total pixel size of 13.3 MPix and a 'usable' pixel size of 12.1 MPixels , the difference usually comes from the borders where some pixels are not supposed to be used.

Apparently APP/APG displays the source images with the usable pixels, but the panorama is build using the total pixels.

Bernard

mediavets wrote:

slaughter wrote:

I'm using a Canon Powershot s100.
For some reason, when I import the CR2 files directly into Autopano,
the extremeties of the image contain more of the panorama than when I convert them to TIFF first.  It's very strange.  The CR2 panorama literally contains more of the landscape, it's a wider image. The TIFF version is literally smaller.  I'll upload a screenshot after work to ilustrate.
Anyway, this is why I want to be able to do the whole thing in Autopano - and I have no idea where the colour information is going.  I can't find any settings to fix the problem.

Apparently it's an issue with dcraw:
http://code.google.com/p/photivo/issues/detail?id=71

"the culprit is not Photivo but dcraw. Photivo has dcraw 1.445 built-in to decode raw files. Standalone dcraw 1.445 exhibits the same problem. The good news: The most recent standalone dcraw 1.447 works fine."

APP/APG RAW conversion also uses dcraw.

Maaannnn.  I even downloaded the latest, which is dcraw 1.5.6 (http://dcraw-x.mac.findmysoft.com/download/#) for mac, but the software *doesn't* recognise CR2 files or support the Canon s100.   How is it that version 1.446 would work...? surely the latest version would be a more complete version...?

Looks like I might get all the useable pixels happening if I convert to TIFF using dcraw first, but I can't find any dcraw apps which will recognise CR2 files.

mediavets wrote:

slaughter wrote:

A final look at this:  i have also attached how APG sees the JPG or TIFF of a single image, vs how APG sees the CR2 version of the same image.  The darned thing is BIGGER. WHY? When I view the CR2 in Canon DPP, OR in PS, the image is as per the colour one.  Yet when I import the CR2 into APG, APG magically sees about 10% more photograph than what ANY other software will show me.  Help!  The nightmare of this is APG saying "check it out, I will create a panorama containing, magically, more of the landscape you thought was there!  But I'm only going to let you see it in PURPLE!"

It seems that many CR2 RAW convertors for S100 apply geometric distortion corrections - as apparently does the camera when processing JPEGS. That probably accounts for the difference in FOV that you describe and that one can see in the two sample images/screenshots you posted when there's obviously some distortion correction going on in the JPEG.

http://forums.dpreview.com/forums/readf … angemode=1

Understood. So where can I find software that *won't* apply corrections and *will* show me ALL useable pixels, like APG evidently does?  Someone mentioned dcraw 1.447 but version 1.556 doesn't even recognise CR2 files.  I can restart my mac as a PC, is there some PC software that will do this conversion properly for me...?

DXO seems to convert to TIFF using all available pixels -

According to the output *data* (ie what the inspector window tells me) for both versions, the DXO TIFF file is in theory a more detailed and information rich document.  Yet when I look at it, it just doesn't look as good to me.  Please look at the comparison images - the first, arguably granier image, is converted via Camera RAW 6.6 in PS (The result is exactly the same if I use the Canon Digital Prof Photo software, incidentally).  The second image is the DXO conversion.  It looks more like a JPG.  I'd be happy to just use Camera RAW or DPP instead, however DXO is the only application I have found so far which uses ALL of the available pixels in the CR2 file, meaning I can output a TIFF of a larger, more expansive landscape than If I use Camera RAW or DPP.  The Camera RAW conversion approximates more closely the look and feel of film to me.  This is how I want DXO to convert my CR2, as well as the additional pixel data.

*for some reason, the hyperlinks lead to dead ends - you'll need to copy and paste the URLS into a new browser window to see the images!*

Photoshop (Camera RAW 6.6) CR2 to TIFF conversion:

http://6pix.net/images/77923070326923293733.png

36mb
4000x3000
Color Model: RGB
Depth: 8
DPI Height: 240
DPI Width: 240
Pixel Height: 3000
Pixel Width: 4000

DXO CR2 to TIFF conversion:

http://6pix.net/images/57879587856708628862.png

72.2mb
4000x3000
Color Model: RGB
Depth: 16
DPI Height: 300
DPI Width: 300
Pixel Height: 3000
Pixel Width: 4000

So basically I'm almost there but not quite.  Any suggestions would be appreciated.

**I've since established that the program by default applies a lot of "corrections", so I have turned ALL of them off - including those in the advanced workspace. 

Thing is, the damned thing STILL shows a "correction preview" which has very slightly softened the image.  When I click on the image, to show "as shot", it looks as it should, nice and grainy, when I unclick, it is slightly softened as "correction preview". Yet I've turned ALL corrections and alterations off!

Isn't there a simple way to export with NO CORRECTIONS, "AS SHOT" ?

I can't find such a setting anywhere.**

Last edited by slaughter (2012-05-09 15:54:53)