Pages: [1]   Go Down

Author Topic: PercepTool experiences?  (Read 8398 times)

feppe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2906
  • Oh this shows up in here!
    • Harri Jahkola Photography
PercepTool experiences?
« on: May 31, 2009, 04:44:20 am »

There's a new PS plugin called PercepTool which promises a lot. Here's the marketing blurb:

Quote
PercepTool takes the image made in the digital camera sensor (also called luminance image) and changes it into what our brain actually perceives. The luminance image consists of 2 sub (intrinsic) images: reflection and illumination. The reflection and illumination images are separated from the luminance image in the visual cortex of the brain, processed and recombined into what is known as the percept, or the image that we actually perceive. We call this image luminosity. Most of the changes are due to differential recombination of the edges in the image and optimization of tonal values.

I'm highly skeptical of this as the examples don't look anything beyond basic curves adjustment. And even more importantly, it is technically impossible to decompose the RGB value of a pixel to its reflection and illumination value, as is implied in the quote above, so the whole premise is highly suspect. Not to mention the plugin has taken longer to make than the Pyramids.

But I'm open to being proven wrong. It is currently available for Mac only - has anyone tried the trial or the full version, and could give insight or examples?

hsmeets

  • Full Member
  • ***
  • Offline Offline
  • Posts: 184
PercepTool experiences?
« Reply #1 on: May 31, 2009, 07:42:12 am »

As it has a demo mode.....I gave it a try.

[attachment=14154:perceptool1.jpg]

Is the original file as scanned from the negative.

[attachment=14155:perceptool2.jpg]

Is the file but then put through the Plugin. While the dialog is open, there are two extra adjustment layers visible: Gamma and Saturation. Between these is the layer called 'Perceptool' which seems to be a blending layer with grey values calculated from the original image, maybe an attempt to improve 'local contrast'. There is a hotspot visible.

[attachment=14156:perceptool3.jpg]

This is my own interpretation with some selected curves adjustments...

I tried it on a few other images but I'm not impressed.

Logged
Cheers,

Huib

pcmurray

  • Newbie
  • *
  • Offline Offline
  • Posts: 12
PercepTool experiences?
« Reply #2 on: May 31, 2009, 08:37:26 am »

DPreview has 121 posts on one thread...almost all negative.
http://forums.dpreview.com/forums/readflat...6486&page=1

 As stated before on another thread here, I found the book and demo severely wanting. The hype just didn't produce results. The book is heavy on marketing and the software is a hefty price. I found no important information from the book . Oh, Mr DeWolfe wrote that my one click on his software took him 30 years to develop.  But the demo for my mac did not work at all (quad core and got a run time error after waiting five minutes). So I am out the price of the book and removed the software from my system.

Regards,
pc murray
« Last Edit: May 31, 2009, 11:08:47 am by pcmurray »
Logged

Geoff Wittig

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1023
PercepTool experiences?
« Reply #3 on: May 31, 2009, 09:54:50 am »

Quote from: feppe
There's a new PS plugin called PercepTool which promises a lot. Here's the marketing blurb:



I'm highly skeptical of this as the examples don't look anything beyond basic curves adjustment. And even more importantly, it is technically impossible to decompose the RGB value of a pixel to its reflection and illumination value, as is implied in the quote above, so the whole premise is highly suspect. Not to mention the plugin has taken longer to make than the Pyramids.

But I'm open to being proven wrong. It is currently available for Mac only - has anyone tried the trial or the full version, and could give insight or examples?

1) George DeWolfe I think represents the current incarnation of the mystical/metaphysical approach to landscape photography, linked in the past to folks like Minor White (not coincidentally his mentor) and Wynn Bullock. You don't have to buy into the mysticism to enjoy their work, if it appeals to you. I really love David Vestal's thumbnail assessment of Minor White: good man, great photographer, shame about the nutty philosophy but the photos are still great. Some of George DeWolfe's work is very nice; I don't personally buy his nebulous mysticism, but there's something to his notion of 'presence' in a print. There are definitely times when a relatively minor adjustment of tonal distribution or shadow density suddenly permits a print to come to life.
After reading DeWolfe's book I can see there are a few decent ideas there, but overall nothing earth shattering. I much prefer Charlie Cramer's pragmatic and concrete approach to Photoshop and artistic image editing. Cramer teaches a fantastic Photoshop image editing/printing course that comes with a really excellent and concise spiral-bound guide; I really hope he comes out with a book.

2) As a physician I have a nodding familiarity with models of visual perception, and they are just that: models. To this day no one really knows precisely how visual information is collated and processed. Mr. DeWolfe's model is a gross oversimplifcation that leaves out multiple levels of processing, including a very sophisticated processing step that occurs within the retina and its associated nerve fiber layer. There is also profound modification of visual processing by pre-existing visual memory and expectations. Galen Rowel actually spent years studying issues of perception, and wrote extensively about it with a level of nuance and sophistication considerably beyond DeWolfe's.

3) I'm inclined to cut Mr. DeWolfe some slack for the flowery and over-the-top claims he makes for his software; after all, he's selling a product, so you have to expect him to talk it up a bit. And there are lots of plug-ins and programs that are one-trick-ponies yet cost even more. $250 for Helicon focus comes to mind. But he does seem more than a little "out there", particularly considering how slow and kludgy the product has turned out to be. A little modesty might be refreshing; you never know.
Logged

JeffKohn

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1668
    • http://jeffk-photo.typepad.com
PercepTool experiences?
« Reply #4 on: May 31, 2009, 11:41:53 am »

I've been playing with the Windows trial for a couple of days. The plugin is painfully slow, and has completely crashed Photoshop several times (running the 32-bit version of CS4 on Vista x64).

The plugin itself seems to slightly darken bright areas, slightly boost shadow areas, and increase local contrast. For some images it works fairly well, for some images it doesn't have much effect, and for others it will make at least part of the image worse and not better. The gamma and saturation sliders don't really have anything to do with PercepTool itself, if you look at the layers stack for your image while the dialog is up you'll see that it's just using  Expsoure and Hue/Saturation adjustment layers (hint, open the Adjustments palette in CS4 before running the PercepTool script and you'll see). The script UI with the sliders is also pretty crude/buggy.

There's no magic here, and it certainly doesn't translate the image into what my brain perceived when shooting. This plug-in offers a mild improvement for some images (if you can suffer the interminable wait time), but doesn't deliver nearly as much as the marketing hype promises.
Logged
Jeff Kohn
[url=http://ww

luong

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 259
    • http://www.terragalleria.com
PercepTool experiences?
« Reply #5 on: June 05, 2009, 07:21:51 pm »

Quote from: feppe
it is technically impossible to decompose the RGB value of a pixel to its reflection and illumination value, as is implied in the quote above

Yes, in general the problem *is* clearly  ambiguous. If you make some assumptions about the "regularity" of the world, it is possible to come up with *plausible* answers, however it is an extremely difficult problem that has been attacked by top-notch perception/artificial intelligence researchers without  entirely satisfying solutions, so one should not expect miracles from a PS plugin written by a photographer. Indeed I agree that this reference makes the software somehow suspect.
Logged
QT Luong - author of http://TreasuredLandsBook.com, winner of 6 national book awards
Pages: [1]   Go Up