[APG 3.x Windows XP/32] APG 3.x and PTGui Pro ....

[mediavets]

by mediavets » Wed Aug 07, 2013 9:55 am

That would not be entirely possible.. The VR Drive HDR software works on the rang of the cameras and also I have set my ISO to 100, which is not possible with my D90....... Also, I do not think HDR will work with my D90 since it requires the use of a USB cable...

Destiny...

It seems that the D90 is compatible with HDR mode:

http://www.roundshot.ch/xml_1/internet/en/application/d77/d78/f86.cfm

I think the USB cable may be 'universal'?

The same ISO setting is not an issue really. 'Similar' workflow is good enough.

What we are exploring is the APG 3.x sticthing and rendering of bracketed image sets exposure-fused with Photomatix.

[Artisan New]

by Artisan New » Wed Aug 07, 2013 10:22 am

Not in the world of Nikon.....Nikon actually has "managed" to use three different types of release connectors

http://www.doc-diy.net/photo/remote_pinout/

being:

MC-DC1, MC-DC2 (for the D90) and the MC-20, MC-22, MC-30 and MC-36.....amongst which lurks the D800....

and in USB cables they managed to "invent" about 15 different types of "Universal" serial bus cables...being:

UC-E1,UC-E2,UC-E3,UC-E4 (used for the D90), USB-E5, USB-E6, USB-E10, USB-E12,ESB-E13, USB-E14 (used for the D800), USB-E15, USB-E16, USB-E17, and USB-E19....I say developers like that should be banned from ever selling one product again.

https://nikoneurope-nl.custhelp.com/app/answers/detail/a_id/10166/~/welke-usb-kabel-van-nikon-heb-ik-nodig-voor-mijn-digitale-camera%3F

Greets, Ed.

P.S, your D90 can be used according to the site of Seitz even HDR...

http://www.roundshot.ch/xml_1/internet/en/application/d77/d78/f86.cfm

Last edited by Artisan New on Wed Aug 07, 2013 10:28 am, edited 1 time in total.

[AlexandreJ]

by AlexandreJ » Wed Aug 07, 2013 10:28 am

Sorry I want to play with the sample too, but I didn't found any link to the image set. Where can I find it ?

[Artisan New]

by Artisan New » Wed Aug 07, 2013 10:30 am

Eh, Des's or mine.....

[AlexandreJ]

by AlexandreJ » Wed Aug 07, 2013 10:55 am

I found Destiny sample. Playing with it right now.

[Destiny]

by Destiny » Wed Aug 07, 2013 11:09 am

Alex, make sure you use the current set.. from my Dropbox... If you need a new link I will send it to you...

Destiny..

I found Destiny sample. Playing with it right now.

[Destiny]

by Destiny » Wed Aug 07, 2013 11:18 am

Ok.. I will look into it more but I cannot use the same USB connection... The D800 uses the USB 3.0 A to Micro-B Cables where as the D90 might use the USB 2.0 A to Mini-B, but I will have to check...

Destiny...

That would not be entirely possible.. The VR Drive HDR software works on the rang of the cameras and also I have set my ISO to 100, which is not possible with my D90....... Also, I do not think HDR will work with my D90 since it requires the use of a USB cable...

Destiny...

It seems that the D90 is compatible with HDR mode:

http://www.roundshot.ch/xml_1/internet/en/application/d77/d78/f86.cfm

I think the USB cable may be 'universal'?

The same ISO setting is not an issue really. 'Similar' workflow is good enough.

What we are exploring is the APG 3.x sticthing and rendering of bracketed image sets exposure-fused with Photomatix.

[lumelix]

by lumelix » Wed Aug 07, 2013 12:00 pm

You're a lucky guy Martin ...but indeed did you use Photomatix?

I'm using this 16mm since 20 years and happy how good it performs on the D800. Super sharp at f8 to f16 and no flares.
So I prefer this lens when shooting into the sun.

I do a lot of HDR with this combination. But always use a strong tripod. So normaly I mustn't use the align-function in Photomatix.
This function can make you trouble because it cuts of the empty edges so you get different image sizes. Perhaps this can confuse APG ?
I also use only every second shot of the bracketing (eg: -5, -3, -1, +1, +3 EV). It's absolutely enough to avoid noise and to get a great dynamic.

Hope you find a solution for this problem.

[AlexandreJ]

by AlexandreJ » Wed Aug 07, 2013 12:23 pm

I did play a lot with Destiny sample and made an analysis only on the geometric registration (color / antighost decision were not tested).
I used Autopano Engine in v3.0.7 (either Pro or Giga, both works the same way).

Here my workflow and how I found out the issue with this case.
- default settings: rendering at full scale. Some issue along the ceiling where the lines are broken.
- I check with cp editor, everything seem fine without any real issue. I switch focal length option, from "auto" to "force different" and lens distortion from "auto" to "force same".
- Reoptimize using quick optim. 2, 3 times until the global RMS didn't change anymore.
- Render again, look at full scale and the ceiling lines are not broken now.

So, why should I need to change focal length to force different ?

It seems that before autopano, as photomatix was used, there might be some registration been done to adjust this stack that could change a little the focal length. And as this operation is not being done the same way on all stacks, the real resulting focal length coming out of photomatix could be bit different between the images we have in input. Or autopano is handling focal based on exif. Here, the focal length length in the exif is just the same, so autopano solves for one focal for the whole project and not one per image.
This is what I did tell to autopano when changing the optimization options. Why did I force lens distortion to "force the same" ? Because in auto mode, it follows the focal length, so if I didn't changed it, it would have search for lens distortion per image instead of per project.

I did check further. If this is real explanation, just change this setting in the group properties and run the detect to get a perfect geometric stitch. This doesn't work and I don't know why. It seems to like the fact that detection run in default and then edit, change the option, optimize and then it is perfect. I have to look further into why It didn't work directly.

Why did it work nicely with 2.0.9 ?

Probably because the lens distortion model is far more simple and so the optimization process doesn't stuck itself in a false solution (2.0.9: for fisheye, 2nd order, no offset). Or because of Smartblend that can fix quite well broken lines. I did check that model with 3.0.7 and it is really bad in registration quality.

Cleaning control point. Needed or not ?

In optimization, advanced settings, you can uncheck "cleaning control point". In that case, you'll get all control points even some less good ones.
I did that with this case, everything else in default mode, detect, change the focal length option, reoptimise (ended RMS 5.2), render at full scale. It was also perfect.
So not bad control points doesn't really influence here as globally the goods one are outnumbering the not so good one.

Even better lens model. k3 activation needed ?

In the perfect stitch, you could see a half pixel shift in one line. Activating this 3rd order distortion did remove that. But global RMS didn't change when hitting optimization. I hardly think it is needed.

That's my current finding on Destiny case.

[klausesser]

by klausesser » Wed Aug 07, 2013 12:27 pm

You're a lucky guy Martin ...but indeed did you use Photomatix?

I'm using this 16mm since 20 years and happy how good it performs on the D800. Super sharp at f8 to f16 and no flares.
So I prefer this lens when shooting into the sun.

I do a lot of HDR with this combination. But always use a strong tripod. So normaly I mustn't use the align-function in Photomatix.
This function can make you trouble because it cuts of the empty edges so you get different image sizes. Perhaps this can confuse APG ?
I also use only every second shot of the bracketing (eg: -5, -3, -1, +1, +3 EV). It's absolutely enough to avoid noise and to get a great dynamic.

Hey Martin!

The 16mm FE is a great lens, yes (as is the 15mm Canon FE btw.) I compared it to the somewhat ledendary 2,8/8mm FE (which is hard to get these days and very expensive if you find a good one and it´s extremely delicate and vulnerable) and the 8mm had a hard time. I was close to buy one last year - but thanks heaven i had the chance to compare them first . . Fine lens for collectors nevertheless: http://www.kenrockwell.com/nikon/8mm.htm

Using the "align images" option in Photomatix should never be used with images to be stitched! That´s obvious. PTGui for example usually refused to take them at all (never tried it again - maybe it changed actually).

The bracketing-options is the only feature which REALLY leaves me unsatisfied with the D800. I use PromoteControl for that purpose - that one does MUCH better with bracketing (after a firmware upgrade).

best, Klaus

Last edited by klausesser on Wed Aug 07, 2013 12:46 pm, edited 1 time in total.

[Destiny]

by Destiny » Wed Aug 07, 2013 12:41 pm

Wow Alex.. that was some testing.. Thank you. Its made me feel a lot better to realise its not me failing to understand how APG worked, and not knowing what I am doing.. It will make me very happy to see the shine come back into APG for me.. I have tried very hard to get APG to work with my images.. What ever work flow you find that works, I will follow it.. I will have to try this tomorrow.. Run out of time again ;(

Destiny..

[mediavets]

by mediavets » Wed Aug 07, 2013 12:46 pm

I did play a lot with Destiny sample and made an analysis only on the geometric registration (color / antighost decision were not tested).
I used Autopano Engine in v3.0.7 (either Pro or Giga, both works the same way)............

So, why should I need to change focal length to force different ?

It seems that before autopano, as photomatix was used, there might be some registration been done to adjust this stack that could change a little the focal length. And as this operation is not being done the same way on all stacks, the real resulting focal length coming out of photomatix could be bit different between the images we have in input. Or autopano is handling focal based on exif. Here, the focal length length in the exif is just the same, so autopano solves for one focal for the whole project and not one per image.

Alex,

Thanks for investigating this case which is just one example of a bracketed exposure image set fusion pre-processed using Photomatix.

You offer some speculations about the possible influence/impact of pre-processing with Photomatix on stitching precision using APG 3.0.7; and offered a possible explanation for why APG 2.0.9 was able to produce a prefect stitch when APG 3.0.7 didn't.

If your speculations are relevant why do they not also affect PTGui Pro which produced a perfect stitch when I chose the option to use the EXIF data to parameterise lens data?

I am pleased to hear you will be undertaking further investigations to try to enable a simpler remedial workflow and look forward to hearing of the results of that.

.............

I feel that it's important for APP/APG to be able handle bracketed exposure image sets pre-processed with Photomatix successfully and easily; and not only because PTGui Pro seems to do it every well.

Photomatix is a very popular and well regarded program for exposure fusion and HDR, and used in combination with APP/APG enables people to handle large bracketed exposure image sets on much lower pperfomance hardware that is required if attempting exposure fusion within APG; and hardware costs are an important factor for many hobbyists who (unlike pros) cannot easily justify constantly upgrading to the latest and greatest systems.

Last edited by mediavets on Wed Aug 07, 2013 12:47 pm, edited 1 time in total.

[klausesser]

by klausesser » Wed Aug 07, 2013 1:02 pm

You offer some speculations about the possible influence/impact of pre-processing with Photomatix on stitching precision using APG 3.0.7; and offered a possible explanation for why APG 2.0.9 was able to produce a prefect stitch when APG 3.0.7 didn't.

i asked around some other pamorama-photographers: not a single one of them, including me, ever had an issue stitching Photomatix-processed images in APG.
So i guess it´s a question of some individual settings - i suspect the reason might be the use of Photomatix´ align-feature . . which i - and the others - never use with images to be stitched.

If your speculations are relevant why do they not also affect PTGui Pro which produced a perfect stitch when I chose the option to use the EXIF data to parameterise lens data?

In fact it does: using the "align" feature in Photomatix PTGui then refused to take the images at all saying: "all images must have the same size". At least i encountered this in an earlier version - and never again used "align" in Photomatix. . .
Maybe that´s why i - and the others - never had any issue stitching pre-processed images

I fele that it's important for APP/APG to be able handle bracketed exposure image sets pre-processed with Photomatix successfully and easily; and not only because PTGui Pro seems to do it every well.

Photomatix is a very popular and well regarded program for exposure fusion and HDR, and used in combination with APP/APG enables people to handle large bracketed exposure image sets on much lower pperfomance hardware that is required if attempting exposure fusion within APG; and hardware costs are an important factor for many hobbyists who (unlike pros) cannot easily justify constantly upgrading to the latest and greatest systems.

Again: i don´t think this is a basic issue in Photomatix or APG but rises with choosing a wrong setting in Photomatix.
Should be investigated by Photomatix as well as by Kolor nevertheless.

best, Klaus

[mediavets]

by mediavets » Wed Aug 07, 2013 2:17 pm

I did play a lot with Destiny sample and made an analysis only on the geometric registration (color / antighost decision were not tested).
I used Autopano Engine in v3.0.7 (either Pro or Giga, both works the same way).

Here my workflow and how I found out the issue with this case.
- default settings: rendering at full scale. Some issue along the ceiling where the lines are broken.
- I check with cp editor, everything seem fine without any real issue. I switch focal length option, from "auto" to "force different" and lens distortion from "auto" to "force same".
- Reoptimize using quick optim. 2, 3 times until the global RMS didn't change anymore.
- Render again, look at full scale and the ceiling lines are not broken now.

Alex,

Hmmmm.....

Your workaround worked well with one test image set - image 1.

But not so well with a second image set - images 2 and 3.

When I switch the focal length and lens distortion setting should I be doing this at the Group level or the Global level?

Last edited by mediavets on Wed Aug 07, 2013 3:21 pm, edited 1 time in total.

[Artisan New]

by Artisan New » Thu Aug 08, 2013 9:27 am

i asked around some other pamorama-photographers: not a single one of them, including me, ever had an issue stitching Photomatix-processed images in APG.
So i guess it´s a question of some individual settings - i suspect the reason might be the use of Photomatix´ align-feature . . which i - and the others - never use with images to be stitched.

I think you are right Klaus (this hurts not as much as I would think it would hurt )....i've found the problem in MY usage of Photomatix/Autopano Pro....

1) I've use default settings in Photomatix in order to merge the brackets....now that is not the point of using Photomatix. Some mailing back and forth and some reading (some HDR literature online and offline) later I redone my files.....they were looking a lot better and they were more appreciated by Autopano Pro....

2) I used the setting recommended by Alexandre.....now what did the trick remains to be investigated....but exposure fusion now worked.

Greets, Ed.

P.S. that saves me 150 euro, so Klaus I ow you a bear or two....

[Artisan New]

by Artisan New » Thu Aug 08, 2013 9:39 am

Carefull, Klaus, this is camera porn.....

http://www.youtube.com/watch?v=zIDbw4gjunY

Nikon also has a 6mm 2,8.....

Greets, Ed.

[klausesser]

by klausesser » Thu Aug 08, 2013 10:58 am

Nikon also has a 6mm 2,8.....

I did two jobs using this lens - i rented it in New York back in 1971. It cost a fortune to buy.
Long time out of production you find it exremely rarely theses days for the cost of a luxory car.

http://www.mir.com.my/rb/photography/companies/nikon/nikkoresources/fisheyes/6mmf28.htm

Klaus

Last edited by klausesser on Thu Aug 08, 2013 11:00 am, edited 1 time in total.

[marzipano]

by marzipano » Thu Aug 08, 2013 3:42 pm

I do a lot of HDR with this combination. But always use a strong tripod. So normaly I mustn't use the align-function in Photomatix.
This function can make you trouble because it cuts of the empty edges so you get different image sizes. Perhaps this can confuse APG ?
I also use only every second shot of the bracketing (eg: -5, -3, -1, +1, +3 EV). It's absolutely enough to avoid noise and to get a great dynamic.

Hope you find a solution for this problem.

As I do quite a number of panoramas hand-held with bracketing, the Alignment facility in Photomatix is important to enable slight (or sometimes worse)camera movements to be eliminated in the final image

It is when it is used with the "Crop" check box ticked that the photomatix images to be stitched in APG can all end up slightly different sizes. Without checking the "crop" box, then they all end up as the original input size but with white borders round the unmatched portions which is also confusing to APG

However, I have noticed that if you don't select any alignment but do select anti-ghosting plus "automatic" ghost removal (not "manual" selection of specific areas), then this would also appear to carry out alignment of bracketed images in addition to anti-ghosting and also keeps the original dimensions intact and does not produce a white border round the ummatched sections (i.e exactly what APG wants)

The Photomatix manual does not, however, back this up so it may be an unintented feature but it seems to work quite reliably as far as I can tell (I'm on v4.2 Pro and using Fusion with the "Natural" settings)

[klausesser]

by klausesser » Thu Aug 08, 2013 4:14 pm

However, I have noticed that if you don't select any alignment but do select anti-ghosting plus "automatic" ghost removal (not "manual" selection of specific areas), then this would also appear to carry out alignment of bracketed images in addition to anti-ghosting and also keeps the original dimensions intact and does not produce a white border round the ummatched sections (i.e exactly what APG wants)

Very good point! I always have it activatd - cleans up many things.

Another clue with "auto ghost removal": if you have moving flames in the brackets - torches for example - you´ll most likely have some strange looking parts around the flame after processing.

Activating "auto ghost removal" in most cases cleans these areas up.

best, Klaus