Landscape and stitching

[Asher Kelman]

Hi Bart

glad you jump in. For downrez to native size, I used your Gausschenblur-methode.

What is that, beside the obvious translation!

yes, they' re from the 1 Ds-2; with 9952 x 6624 (in LR, ACR, and RAW Developer)
iVMP looks at them at 10004 x 6656!

which Photoacute-version do you use? It's 2.7.2, here.

Edit: It was done with HDR-methode (at the top), with 2x Superes + HDR at the center....
3 brackete RAWs, + 1.5, 0, - 1.5...

Michael,

Are you saying you mix bracketed exposures with single exposure in one Pano? Is there a special order or else one can dump in a whole folder with everything, multiple exposures of one part, other areas bracketed exposures etc? It seems you must be talking about using PhotoAcute with another Pano program, but I'm not clear on what you are doing.

Also, what do you use to open your huge files? What works and what doesn't?

Asher

 

[Michael Fontana]

Bart

actually I' ve 2.7.3, here.
As PA has a scratch-architecture, maybe there' s the bug; as here it sucks approx 500 MB physical RAM & 700 MB of virtual memory only, for 3 raws. Meanwhile scratch is about 900 MB.... for 1 x size, when trying focus-stacking.

Asher

Bart's downrez-hint is post Nr 7, in here

Really usefull, as PS in bicubic smoother/sharper produces jaggie's.

Panos+HDR:

Before PhotoAcute, I went the Photomatix-track = creating tiffs out of the brackets RAWs, HDRing and tonemapping them in Photomatix, exporting as tiffs and and stitching them in PTGUi.
PTGui has a HDRfunction in the Proversion, but I don't like its output at all.

As PhotoAcute has a HDR-function as well, which showed some better definition than Photomatix, in some single = normal shots, I redid the entire pano from scratch, using the RAWs:

Sending the 3 bracket RAWs to PhotoAcute, - using its HDR-function plus the supperresolution, in the setting " High Dyanmic Range, tripod mounted" plus the sub- setting " Combined 2 x output, with superresolution" tries to squeeze every little detail out of the RAWs, prior to convert it to tiffs. = "Pump-up methode".


Then, exporting as DNG and converting them in LR to tiffs. For a pano, this has to be done for each lens position = pano section. The tiffs out of LR were stitched with PTGui.

The examples from page 3, post 25, show a comparison of the PhotoAcute-workflow vs the Photomatix-one, using identical RAWs, with 3 diff. exposures, - 1.5 f, 0, + 1.5 f. All the screenshots are from the identical mountain pano, but HDRing/tonemapping been done with 2 different tools.
Is it understandable now?

There's a substanceable gain in definition with Photoacute, giving the panos more presence, less noise, etc.

 

[Asher Kelman]

Thanks so much for the detail Michael!

I'd love to be able to automate this. It should be possible in OSX!

I'm using CS2 and keep hitting the 4GB barrier in opening the Pano files from Autopano Pro. Wat's the limit on your various final software packages?

The time taken to process files is too much. I'm hoping for a major improvement in new computers. We may need to be using some different architecture!

Otherwise 2 sheets of film scanned and combined in Photoacute might be a faster route!

Asher

 

[nicolas claris]

Thanks so much for the detail Michael!

I'd love to be able to automate this. It should be possible in OSX!

I'm using CS2 and keep hitting the 4GB barrier in opening the Pano files from Autopano Pro. Wat's the limit on your various final software packages?

The time taken to process files is too much. I'm hoping for a major improvement in new computers. We may need to be using some different architecture!

Otherwise 2 sheets of film scanned and combined in Photoacute might be a faster route!

Asher

Hi Asher
If you are on Macpro or G5, CS3 is muuuuuuuuuuch faster…

;-)

 

[Asher Kelman]

Hi Asher
If you are on Macpro or G5, CS3 is muuuuuuuuuuch faster…

;-)

Thanks nicolas!

For 4GB files? What's the highest size you can open or save in CS3?

My workflow block is in assembling Pano's using AutoPano Pro on my G5.

I'll try APP in my MacBookPro?

I'm wondering about limits in GIMP?

Asher

 

[nicolas claris]

Sorry, I was speaking of speed of production opening, saving, filtering, masking, cloning etc…

I don't rmember having worked on greater file then 3 Gb, I'll give it a try tomorrow with CS3 (brand new CS3, Michael;-) on my Macpro with its 8 GB of ram and let you know…

BTW, it's really a pitty that CS3 (as previous versions) cannot use more than 4 GB of ram…

 

[Asher Kelman]

Nicolas,

OS X may recruit more RAM for Photoshop than that 4GB that PS admits!

For opening files in CS3, could you try a greater than a 4 GB file?

Asher

 

[Michael Fontana]

..........
I'd love to be able to automate this. It should be possible in OSX!

I'm using CS2 and keep hitting the 4GB barrier in opening the Pano files from Autopano Pro. Wat's the limit on your various final software packages?

The time taken to process files is too much. I'm hoping for a major improvement in new computers. We may need to be using some different architecture!

Otherwise 2 sheets of film scanned and combined in Photoacute might be a faster route!

Asher

Asher,
I' ve been asking PhotoAcute's support for a batch-possibility, already.

> 4 GB with PS is not really fast, but possible. I had about a 7 GB-files (psd) already, but not with stitching, though...

When stitching big, you best use saving as > .psb< ; that should help. Probably you don't hit the PS 4 GB-barrier, but the tiff-limitation, as PS Cs-2 can °suck° RAM from the OS:

"TIFF uses 4-byte integer file offsets to store image data, with the consequence that a TIFF file cannot have more than 4 Gigabytes of raster data (and some files have begun to approach this boundary). However, this is 4G of compressed data, and so if the compression ratio is high enough, theoretically a TIFF image could be much larger (in fact, 2**32-1 pixels square)."

>Otherwise 2 sheets of film scanned and combined in Photoacute might be a faster route!<

I've been thinking about that, but I'm not sure!
One problem would be the alignement of the films on a very precise level, for HDRing...
I still remember the film sheets beeing differently curved.

Carrying a 4/5, with all these films, - folders and lenses (compared to 13 GB of RAWfiles, on that alp-07-shooting) would have been impossible, just for the weight of equipement!

When wanting images with bigger angle's than 90 degr. stitching with zylindrical projection is far superior to any 4/5! You might noticed, that the first two images of that thread have a HFOV of 181, resp. 139 degrees!


And I like much the versality and flexibility with stitching; as the photographer chooses the VFOV and HFOV, not the lens construction.....

if you want - on the spot - add a little bit of widths, no problem, just another shot and you' re done.

 

[Michael Fontana]

Nicolas....>BTW, it's really a pitty that CS3 (as previous versions) cannot use more than 4 GB of ram…<

Look here

I noticed it - with 8.5 GB of RAM on a PPC-Quad - when observing the mac's own utility °activity monitor°.

 

[Michael Fontana]

I gave the PhotoAcute's focus stacking a try: Depth of field expansion, as I might use it in architecture modell' s shots:

A 50 mm-lens has - without focus stacking - a DOF of 4-6 cm at this range and f-stop, meanwhile PhotoAcute enhaces it to >10 cm.


Stacking of focus at 40 cm, 45 cm, 50 cm, with the macro 50 mm on FF:

You might notice the changes of size, that PA has to make, to get the DOF expanded over the entire range: its going from center towards the outside; a inverted layer at 50 % opacity, in top of the other one (in PS) shows it better.

 

[Bart_van_der_Wolf]

I gave the PhotoAcute's focus stacking a try: Depth of field expansion, as I might use it in architecture modell' s shots:
...
You might notice the changes of size, that PA has to make, to get the DOF expanded over the entire range:

Yes that's usual at close proximity, because changing focus distance also changes magnification which is most noticable close-up and needs to be compensated for. Luckily PhotoAcute uses what seems to be an accurate registration/resampling/morphing method. However, in my tests on a small size version of a(n admittedly difficult case of) focus stacking sequence it made a mess of it (maybe more experimentation with the parameters would help). I expect PA to perform better on relatively straightforward 'deep' images.

Bart

 

[Michael Fontana]

Bart

I had a mess at my first try too, but decided then to try identical focus-steps.

As a photographer (increasing DOF) I first choosed 40, 45, 50, 60, 70 cm, meanwhile a shoot every 5 cm from 40 - 70 makes it way better. Still, with that setup some minor retouches are required; but a DOF being 30 cm is quite impressiv - the distance of 30 cm beeing 1:2!

Edit: obviously, some limits might show off; if someone is overdoing it. PA needs to change magnification to much, then...