Pages: [1] 2   Go Down

Author Topic: Oloneo PhotoEngine feedback  (Read 13637 times)

AntoineC

  • Newbie
  • *
  • Offline Offline
  • Posts: 11
    • Oloneo
Oloneo PhotoEngine feedback
« on: May 01, 2011, 12:47:06 pm »

Hi all!

I was just contacted by a user of this forum recommending me to be part of your community and to discuss directly with the members. So, here I am!

I am the creator and main developer Oloneo PhotoEngine, a professional HDR tool. The software is in beta (actually since last July!). You can download the beta at:
http://www.oloneo.com/en/page/download.html

We have more and more professional photographers using the product for landscape photography.  There is an image gallery on our home page that shows a few images:
http://www.oloneo.com

Please post your comments and suggestions about the beta in this thread. I will be happy to answer to your questions directly on the forum.

Thanks,

Antoine Clappier
Creator of Oloneo PhotoEngine
Logged

Chris_Brown

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 974
  • Smile dammit!
    • Chris Brown Photography
Re: Oloneo PhotoEngine feedback
« Reply #1 on: May 01, 2011, 01:06:28 pm »

Please post your comments and suggestions about the beta in this thread.

The feature set looks promising. I wish there was a Macintosh version, and unfortunately I'm not comfortable running Windows on any of my systems.
Logged
~ CB

AntoineC

  • Newbie
  • *
  • Offline Offline
  • Posts: 11
    • Oloneo
Re: Oloneo PhotoEngine feedback
« Reply #2 on: May 21, 2011, 08:26:03 am »

We plan to create a Mac OS Version.

Thanks,

Antoine
Logged

AntoineC

  • Newbie
  • *
  • Offline Offline
  • Posts: 11
    • Oloneo
Re: Oloneo PhotoEngine feedback
« Reply #3 on: May 21, 2011, 08:31:38 am »

Hi all,

We updated the PhotoEngine beta this morning. The latest version (v1.0.400.298) now includes a Lightroom plug-in and a direct export to Photoshop (or the post processing tool of your choice).

The beta is available for download at:
http://www.oloneo.com/en/page/download.html

Thanks,

Antoine Clappier
Logged

walter.sk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1433
Re: Oloneo PhotoEngine feedback
« Reply #4 on: May 21, 2011, 02:30:47 pm »

I enjoy doing HDR.  Most of my work is concentrated on getting the most natural-looking results, although I sometimes do more "interpretive" work as well.

I have been using the Oloneo HDR software through several versions of the beta, and find it to be exquisite in its ability to produce beautiful and natural-looking results.  I use several other programs (NIK HDR Efex Pro, Photomatix Pro v.4, HDR Expose, and Photoshop CS5's Merge To HDR Pro) and find that Oloneo and HDR Expose rank the highest in terms of ease of getting natural results.
Logged

feppe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2906
  • Oh this shows up in here!
    • Harri Jahkola Photography
Re: Oloneo PhotoEngine feedback
« Reply #5 on: May 21, 2011, 03:21:05 pm »

I enjoy doing HDR.  Most of my work is concentrated on getting the most natural-looking results, although I sometimes do more "interpretive" work as well.

I have been using the Oloneo HDR software through several versions of the beta, and find it to be exquisite in its ability to produce beautiful and natural-looking results.  I use several other programs (NIK HDR Efex Pro, Photomatix Pro v.4, HDR Expose, and Photoshop CS5's Merge To HDR Pro) and find that Oloneo and HDR Expose rank the highest in terms of ease of getting natural results.

How do you get natural results? I tried Oloneo perhaps a year ago, and couldn't get a natural result, and none of the presets worked.

I've tried numerous HDR programs over the years, but none beat manual blending for my needs. I usually take just two shots 4 stops apart and blend them manually in PS which gives the most pleasing results, but would consider going the HDR route as long as I don't have to worry about halos and overcooked results.

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: Oloneo PhotoEngine feedback
« Reply #6 on: May 21, 2011, 04:51:43 pm »

I've tried numerous HDR programs over the years, but none beat manual blending for my needs. I usually take just two shots 4 stops apart and blend them manually in PS which gives the most pleasing results, but would consider going the HDR route as long as I don't have to worry about halos and overcooked results.

Hi Feppe,

I hesitate to react in this thread, but I suggest you try SNS-HDR (its Neutral preset adds zero tonemapping, you can take it from there). Personally I prefer to work from Raw-converted TIFFs, but it can also read Raws. Two shots 4 stops apart is not optimal for HDR or Expoure Blending, mathematically you may get good results with 2 stops apart, but common wisdom rather sticks to someting like 1 or 1.33 stop intervals to avoid potential artifacts.

Cheers,
Bart
« Last Edit: May 22, 2011, 05:29:34 am by BartvanderWolf »
Logged
== If you do what you did, you'll get what you got. ==

feppe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2906
  • Oh this shows up in here!
    • Harri Jahkola Photography
Re: Oloneo PhotoEngine feedback
« Reply #7 on: May 21, 2011, 05:45:52 pm »

Two shots 4 stops apart is not optimal for HDR or Expoure Blending, mathematically you may get good resuts with 2 stops apart, but common wisdom rather sticks to someting like 1 or 1.33 stop intervals to avoid potential artifacts.

I distrust common "wisdom," especially in photography. Do you have some research backing up the claim that 4 stops apart is not optimal for exposure blending? Guillermo Luijk has an excellent article on the topic which is my basis for using 4 stops.

I've found two shots 4 stops apart to be plenty in most situations, and I don't have to tackle with halos, multiple masks or artifcacting. I do take third very fast shot (~0.5 secs) when shooting nighttime cityscapes to capture highlights.

I'll check out SNS-HDR.

AntoineC

  • Newbie
  • *
  • Offline Offline
  • Posts: 11
    • Oloneo
Re: Oloneo PhotoEngine feedback
« Reply #8 on: May 22, 2011, 03:34:14 am »

Hi Feppe,

How do you get natural results? I tried Oloneo perhaps a year ago, and couldn't get a natural result, and none of the presets worked.

I don't think that it was Oloneo PhotoEngine! We have added preset support very recently and the first beta was released less than a year ago.

The beta testers say that they get the most natural results with PhotoEngine and that it is the strongest point of the beta (with speed).

The beta is a 18MB download. Take a minute to download it. I think you will like it.

Thanks,

Antoine
Logged

AntoineC

  • Newbie
  • *
  • Offline Offline
  • Posts: 11
    • Oloneo
Re: Oloneo PhotoEngine feedback
« Reply #9 on: May 22, 2011, 05:20:29 am »

Hi Feppe,

I distrust common "wisdom," especially in photography. Do you have some research backing up the claim that 4 stops apart is not optimal for exposure blending? Guillermo Luijk has an excellent article on the topic which is my basis for using 4 stops.

4 stops apart is quite large. Let me explain:

Exposure compensation value is calculated as the base-2 logarithm of the ratio of the exposures of two photos. In other words, the pixel values in the first photo are 16 times lower than the ones found in the second photo (the inverse of base-2 logarithm is 2 to the power of the number of stops, here: 2^4 = 16).
 
Camera censors have a "sweet spot". Below a certain low value, data is buried in noise. After a certain high value, the censor response to incoming light becomes non-linear.

The goal of HDR is to use the range of values between these two extremes for each of the photos you are assembling. The algorithm attempts to use only the best part of each photo in terms of exposure.

Here is an example:
 - Photo 1: regular exposure
 - Photo 2: under exposed of 4 stops (-4EV)

Assuming 8-bit images (the reasoning is the same with 12 or 14 bits but easier to follow).
 - Total range of pixel values is: [0 255]
 - Useful range is, say: [50 200] (under 50 there is too much noise, after 200, signal becomes non-linear)

For both photos, the algorithm will use the range [50 200].
 - Photo 1: [50 200]
 - Photo 2: [50 200]

Now, the Photo 2 is 16 times darker than Photo 1 (4 stops). Because of that, pixel values of Photo 2 must be multiplied by 16 before assembling with Photo 1:
 - Photo 1: [50 200]
 - Photo 2: [16*50  16*200]

That is:
 - Photo 1: [50 200]
 - Photo 2: [800 3200]

As you can see, the resulting HDR image has values ranging from 50 to 3200. Highlights in Photo 1 are fully recovered. There is anyway a serious issue: where are the values from 200 to 800? Answer: nowhere! There is a large gap between the range [50 200] and [800 3200]!


At this point, you should start guessing why the "2 stops rule" makes sense in HDR photography!

With 2 stops, the exposure ratio is 4 (2 to the power of 2). Let's take the previous example:
 - Photo 1: regular exposure
 - Photo 2: under exposed of 2 stops (-2EV)
 
Ranges after blending:
 - Photo 1: [50 200]
 - Photo 2: [4*50 4*200] = [200 800]

The HDR image has now a range of [50 800] and this time there is not gap from [50 200] to [200 800]. To get the same dynamic range than the first example, you could take three photos 2 stops apart:
 - Photo 1: [50 200]
 - Photo 2: [4*50 4*200] = [200 800]
 - Photo 3: [16*50 16*200] = [800 3200]

You end up with a continuous range of [50 3200] without any gaps.


The previous discussion is pure mathematical reasoning (and I have simplified quite a lot what is really happening in HDR / Exposure Fusion).  Does it really matter to practical HDR photography?

Most of the time, the answer is yes, but not always!

The light in most scenes increases smoothly from the darkest shadows to the brightest highlights. Introducing a gap in the tonal range is not a good idea. You should then use the 2 stops rule.  I even recommend setting the photos 1 stop apart for best results (in that case, there is a tonal overlap between each photo that is beneficial).

In the case of night photography, there are mainly two ranges: very deep shadows and highlights (such as street lights). In this special case, there is a natural gap in light intensities in the real world scene. Using 4 stops can make sense in this situation. I would anyway recommend that you use 2 stops or less for other reasons (such as additional image stacking that will reduce noise but that is another long story!)

Regards,

Antoine Clappier

Logged

feppe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2906
  • Oh this shows up in here!
    • Harri Jahkola Photography
Re: Oloneo PhotoEngine feedback
« Reply #10 on: May 22, 2011, 06:57:38 am »

4 stops apart is quite large. Let me explain:

I was asking specifically about exposure blending, not HDR, so your calculations are irrelevant. When doing exposure blending there's no multiplication resulting in gaps - I'm just combining the two exposures with a tweaked luminance mask. As I mentioned in a previous post, I'd shoot differently if I start doing HDR, as I do appreciate HDR requires different shooting regimen than exposure blending.

Also, I shoot the opposite of your example: ETTR exposure, and +4 EV (not underexposure like you propose) which burns out highlights but lifts detail from shadows for increased DR. I'm having hard time understanding why one would even want to do -4 EV except in extreme circumstances, such as street lights in night shots.

AntoineC

  • Newbie
  • *
  • Offline Offline
  • Posts: 11
    • Oloneo
Re: Oloneo PhotoEngine feedback
« Reply #11 on: May 22, 2011, 10:39:43 am »

Hi Feppe,

The +4EV or -4EV question is a matter of point of view since it is a relative value. In the example above, I could use the Photo 2 as the reference and the Photo 1 as the overexposed photo at +4EV (or use +2 and -2EV respectively or any other combination). Mathematically the results are strictly the same.

You are right. Exposure Fusion is different from regular HDR. There is anyway a blending occurring when merging the two or more photos. With two photos, there are three areas in the resulting photo:
 - Pixels coming only from the Photo 1
 - Pixels coming only from the Photo 2
 - Pixels coming from a blending of Photo 1 and Photo 2

In Exposure Fusion software, the blended part is generally produced by the collapse of two Laplacian Pyramids with varying weights (roughly, a stack of difference images at multiple resolutions). Because of the multi-resolution scheme, the blended area can be large and represent a significant portion of the resulting image.

The blended area is incorrect when the two images are too far apart in term of exposure for reasons similar to the ones described in my previous post (essentially: pixels which are quite unrelated in exposure get mixed together). This error can introduce color shift, noise and halos in the blended portions of the image. The problem is generally much less obvious than with regular HDR.

If you blend the images manually in Photoshop or other tools, the previous issues becomes irrelevant as you decide which parts will get blended.

I am speaking from a purely technical point of view and in the general case. I just want to show there are some mathematical reasons behind the 2 stops approach. My goal is certainly not to show that the "2 stops rule" is the definitive method. I see the "2 stops rule" as a guide for beginners. By using it, photographers can avoid major mistakes in most situations.

With more experience, users discover than the needed number of shots and relative exposure are highly dependent on the photographed scene (and the camera used). Some calls from multiples photos separated of 1EV, some others will work with just two at +-4EV. And some, with only one photo and no editing at all! ;-)

Antoine

« Last Edit: May 22, 2011, 10:42:33 am by AntoineC »
Logged

ejmartin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
Re: Oloneo PhotoEngine feedback
« Reply #12 on: May 22, 2011, 11:03:08 am »

Hi Feppe,

4 stops apart is quite large. Let me explain:
[snip]

The numbers here are quite ridiculous.  First of all, the use of 8-bit example data skews the result by lopping off 4 stops of DR available in today's raw files.

Let's assume we are working in linear gamma (where exposure blending works best).  Then the range 50-200 is only two stops out of the available 8 stops of linear 8-bit data.  So of course there is no overlap between the two images' data if one limits the range used for blending that drastically!

So let's use more realistic numbers: Take 12-bit raw data, and blend the top two stops of the 4 stops more exposed image (which will be used for shadows) with stops 6-8 of the less exposed image.  There is plenty of overlap, since there are 3-4 stops more DR below the blending range in the less exposed image.

Sensor non-linearity is not an issue except perhaps in the top one or two stops near sensor saturation, so irrelevant to the blending range which is 4-6 stops below raw clipping (which could only be the same as sensor saturation at base ISO).  

The 16-fold disparity in the size of the quantization step is also not an issue.  Even at base ISO, SNR never exceeds about 250:1 even for large pixels, and so noise dithers tonal gradations sufficiently that the difference in quantization steps will not be noticeable.

The only issue I see with a 4 stop disparity in exposure is that shot noise will be substantially different between the two images in the tonal ranges they have in common; I don't know whether this would be readily apparent, however.
Logged
emil

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: Oloneo PhotoEngine feedback
« Reply #13 on: May 22, 2011, 11:12:04 am »

The only issue I see with a 4 stop disparity in exposure is that shot noise will be substantially different between the two images in the tonal ranges they have in common; I don't know whether this would be readily apparent, however.

Hi Emil,

That's correct. It also shows as more noisy gradients when the global microcontrast is boosted (which is often the case to improve the look of the images with squashed tonal contrast). That's why I currently stick to 1.33 stops (or 2 at most), but then I sometimes see the noise difference between individual images with just 1/3rd stop exposure difference :(

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

feppe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2906
  • Oh this shows up in here!
    • Harri Jahkola Photography
Re: Oloneo PhotoEngine feedback
« Reply #14 on: May 22, 2011, 11:13:28 am »

That's correct. It also shows as more noisy gradients when the global microcontrast is boosted (which is often the case to improve the look of the images with squashed tonal contrast). That's why I currently stick to 1.33 stops (or 2 at most), but then I sometimes see the noise difference between individual images with just 1/3rd stop exposure difference :(

You can see the different in print or 100% zoom on screen?

ejmartin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
Re: Oloneo PhotoEngine feedback
« Reply #15 on: May 22, 2011, 11:20:13 am »

Perhaps there are details that I'm missing, but naively the only difference between HDR and exposure fusion is the tone curve applied after the exposures are combined.
Logged
emil

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: Oloneo PhotoEngine feedback
« Reply #16 on: May 22, 2011, 11:55:11 am »

You can see the different in print or 100% zoom on screen?

At 100% zoom on screen in a side by side comparison, so relatively insignificant in print (unless blown-up large enough). But I'm also not anal about it, so I settle for 1.33 stop increments in exposure fusion, which passes my seal of approval, but I draw a sharp line at 2 stops.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

AntoineC

  • Newbie
  • *
  • Offline Offline
  • Posts: 11
    • Oloneo
Re: Oloneo PhotoEngine feedback
« Reply #17 on: May 22, 2011, 12:08:13 pm »

Perhaps there are details that I'm missing, but naively the only difference between HDR and exposure fusion is the tone curve applied after the exposures are combined.

Exposure Fusion and HDRI are different algorithms. Actually, they cannot be directly compared! You can compare:
 - Exposure Fusion
 - HDRI generation, followed by Tone Mapping

1- HDRI:
The goal of HDRI is to generate an High Dynamic Range Image from a set of low dynamic ones. When this phase is achieved, you get a high dynamic image with usually 32-bit floating point values per channel. The dynamic range is unchanged at this stage (no tone curve is applied).

If you are interested in the mathematical details, read the original Paul Debevec publication:  Recovering High Dynamic Range Radiance Maps from Photographs
http://www.debevec.org/Research/HDR/debevec-siggraph97.pdf

2 -Tone Mapping:
The goal of Tone Mapping is to bring back the HDR image to low dynamic for display or print. The principle is to compress the dynamic range while retaining the details. There is whole range of methods to achieve that. They fall in two main categories:
 - Global methods (applying the same tone curve to all the pixels)
 - Local methods

With local methods, usually the image is switched to the log domain, the image is split in a detail layer and global layer (this is achieved for instance with a Bilateral filter), the global layer dynamic range is compressed and finally recombined with the detail layer. The end result is a low dynamic image with preserved details.


3- Exposure Fusion
Exposure fusion is an empirical method that compresses the dynamic range in one pass without creating any high dynamic data. The idea is very simple: the algorithm cut the photos in pieces, eliminates the ones with “incorrect” exposures and reassembles a low dynamic image. Mathematically this does not make much sense but it works surprisingly well. Some level of blending must occur between the pieces otherwise obvious seams appear (due to the errors already explained in my previous posts).

A paper on Exposure Fusion:
http://research.edm.uhasselt.be/~tmertens/papers/exposure_fusion_reduced.pdf


Regarding your comment about "quite ridiculous numbers"! I admit that using 8-bit values is an over simplification. I used these values to explain the process without entering in too much detail. If you read HDR papers, you will see that a large part of the original dynamic range is rejected. The blending is done using a weighted sum using hat or Gaussian functions. The [50 250] range is not that far from what is happening inside an HDR/Tone Mapping software.

Antoine Clappier
Oloneo
« Last Edit: May 22, 2011, 02:21:13 pm by AntoineC »
Logged

ejmartin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
Re: Oloneo PhotoEngine feedback
« Reply #18 on: May 22, 2011, 04:36:50 pm »

Sounds like we are talking about different things.  The technique referred to in the Luijk article that feppe referenced is not the same as the one you referenced.
Logged
emil

walter.sk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1433
Re: Oloneo PhotoEngine feedback
« Reply #19 on: May 22, 2011, 05:49:55 pm »

How do you get natural results? I tried Oloneo perhaps a year ago, and couldn't get a natural result, and none of the presets worked.

Here is a jpeg that was reduced from a tiff made in Oloneo of 5 images, 1 1/3 stops apart, with tonemapping aimed at the least obvious effects that would still give me shadow detail without blowing highlights.  One exposure would not capture the range of contrast. This was not processed after, and is not what I consider a finished image.

Logged
Pages: [1] 2   Go Up