RAW Comparisson between RPP and the rest!

Hi guys,

I recently bought a donation based RAW File Processor called Raw Photo Processor, which really kicked out all my other RAW Processors for certain tasks, if not all. RPP tends to sub expose the image a bit, but after you tweak the exposure, you end with a very well exposed picture, with details on shadows and highlights. I guess that RPP manages the file in a more logarithmic way instead linear. Also, colors are much more accurate and realistic.

Here are some examples:

Raw Photo Procesor



Lightroom 2.6 ACR 4.6



Capture NX



Aperture 3



RAW IMAGE: http://www.mediafire.com/?tn0zcjjw2yj

Leo

PS: For the next viewers, please, remit your comments to the topic. It doesn't matter how you see the sky on your backyard.

Leo,

Thanks for this interesting comparison. A related thread by Janet Smth, here is worth reading.

Stuart Rae recently compared HDR processors. In many ways, RAW processing is trying to take advantage of the full dynamic range locked n our files. They may do this simply by having highlight and recovery sliders. What might cripple some processors is the need for a simpler instant previews and speed at the expense of fidelity of number crunching which takes more time. Using rounder of numbers and integers in 14 BIT or even 16 BT calculations might contribute to losses of accuracy and robustness of files. RPP working in 32 BT LAB space and using a compressing algorithm to give film like roll-off for the brightest features, might indeed by the processor to use for individual pictures being developed for the richest result.

You have started off with the popular programs, Adobe Camera RAW & Lightroom, Aperture, Capture NX and the unusually productive Raw Photo Processor by a team of one Californian with the help of his friend, a fussy photographer.

It's great that you have kicked this off. RAW processors are evolving fast and taking advantage of these
opens up amazing possibilities, otherwise untapped, to exploit all the information captures by modern digital sensors.

Could you make available the RAW file by sending it to yourself via yousendit.com and posting the URL so that folk could use this file as the standard for their own processors including

Capture One,

Bibble,

Canon's DPP

Iridient RawDeveloper

Silkypix (free version will not handle A700 files)

LightZone

RawTherapee

ACDSee Pro 2

DxO Optics Pro v5

etc

Asher

Hi, Leo,

Thanks so much for kicking off discussion of this extremely important issue.

I endorse Asher's request regarding your further support of individual investigations in this regard.

By the way, Asher, I think the suggestion to do that via yousendit is very apt. I have used that very successfully to send some very large files to a news organization (archives of multiple high-resolution JPEG files, in that case)

Best regards,

Doug

Hi Leonardo,

Thanks for posting the comparisons. It's interesting to see the differences between raw converters.

Also, colors are much more accurate and realistic.

Click to expand...

To be quite honest, I prefer the colours, especially the sky, produced by Lightroom, and that's something I never thought I'd say because I dislike it intensely.
But then I wasn't there, so can't say what colours are 'correct'.

Regards,

Stuart

Hi Stuart,

You don't need to be there in order to know what about the colors. The sky is the same there, here or any other place around the earth ^^ Also, colors like these never are so saturated. You can even see a difference between pictures in the metal of the wagons, wich one looks more realistic. The other pictures show quite a fake color, don't you feel it? Take a look to the shadows, you will see quite a big difference between these pictures also. All except RPP, are quite dark and even black.

Leonardo Boher said:

Hi Stuart,

You don't need to be there in order to know what about the colors. The sky is the same there, here or any other place around the earth ^^ Also, colors like these never are so saturated. You can even see a difference between pictures in the metal of the wagons, wich one looks more realistic. The other pictures show quite a fake color, don't you feel it? Take a look to the shadows, you will see quite a big difference between these pictures also. All except RPP, are quite dark and even black.

Click to expand...

Hi Martín,

Surely you must realize that it is impossible for anybody to form a good opinion about the quality of this raw converter compared to others based on the jpg images you have presented above? There are millions of factors which stand in the way of forming an informed and semi-objective opinion. I wouldn't want us to head towards a discussion of zealots. Absolute statements presented as the truth do tend to agitate others who might disagree.

And BTW, the sky is not the same at any other place around the earth. Even at the same place it changes every second

Cheers,

Cem_Usakligil said:

Hi Martín,

Surely you must realize that it is impossible for anybody to form a good opinion about the quality of this raw converter compared to others based on the jpg images you have presented above? There are millions of factors which stand in the way of forming an informed and semi-objective opinion. I wouldn't want us to head towards a discussion of zealots. Absolute statements presented as the truth do tend to agitate others who might disagree.

And BTW, the sky is not the same at any other place around the earth. Even at the same place it changes every second

Cheers,

Click to expand...

I have uploaded a RAW File. However, JPEG are enough for comparison, since most people here have not nothing beyond cinema apple displays, like Eizo monitors.

As for the sky, you're right. It doesn't matter how it looked at the end but surely, I have never seen an sky such as the ones shown here, really, but quite enough to say that we can't compare the reality towards digital or even film. We're comparing digital to digital and towards reality, not vice versa.

Here is the same pic, but developed a bit with RPP, specially the exposure, which by default, looks a bit sub-exposed.

..most people here have not nothing beyond cinema apple displays........

Click to expand...

I think you do us a disservice

On my monitor, which is not 'cheap and nasty' and is properly calibrated, the RPP sky looks a bit too grey. Chacun a son gout I suppose.
As for the shadows, most raw converters have a facility to open them out a bit. Have you tried to do this with the alternatives?

Regards,

Stuart

StuartRae said:

I think you do us a disservice

On my monitor, which is not 'cheap and nasty' and is properly calibrated, the RPP sky looks a bit too grey. Chacun a son gout I suppose.
As for the shadows, most raw converters have a facility to open them out a bit. Have you tried to do this with the alternatives?

Regards,

Stuart

Click to expand...

It's about how the raw is interpreted by the app at first sight, without editing.

This isn't realy the best approach though as the whole point of a raw processor is to allow you to control the conversion. It then becomes a question of which processor allows you to achieve what you want in the easiest fashion. I'm no fan of Lightroom - not least because of the way it sprawls across everything you do and squeezes out other options, as well as the poor volour in earlier versions - but you can dramatically change the look of a conversion by some very simple and quite intuitive adjustments. DXo requires you to dig deeper for the ability to change, but gives excellent results with a very little practice.

It can then be quite instructive to compare prints rather than on screen images.

None of this to knock RPP, which I have heard great things about but cannot test as it is Mac only.

Mike

An important aspect of raw data development, about which we rarely hear, is minimization of metameric error. This is a topic I have only come to understand in this past week, and I am now preparing a detailed technical article about it. The study has been very challenging (even expensive, since it has involved the purchase of some international standards I didn't have).

Here is a summary of the matter.

Many different light spectrums can have the same color (which means they look the same color to the eye, and so by definition, they are the same color). This situation is called metamerism.

We look to our sensor to report to us, by way of its three channel outputs, the color of the light. That means that, for any spectrum that "has" that color, the sensor outputs should be the same.

But our sensors generally don't actually achieve that, because of compromises in their design. (In particular, they don't follow the Luther-Ives conditions. This is probably the first time today you have seen those mentioned.) This is called metameric error. It is not a shortcoming that can be corrected in subsequent processing.

When raw data is being developed into an image, about the last step is to transform the color indications from what we can think of as the sensor color space into our output color space (perhaps sRGB). This is done my multiplying the sensor output values by a 3x3 matrix, the result of which is the set of values of the output color space coordinates (in linear form).

But, because of the inconsistency I mentioned above, the sensor doesn't really have a consistent color space. So when we transform from the sensor output coordinates to the output color space coordinates, there is no "transform matrix" that will always "give the right result".

So the designer of the development algorithm uses a matrix designed to minimize the average metameric error over some set of spectrums that are thought to be representative of ones we most often encounter in actual images.

And this can be a difference between the performance of one raw development package and another. As with many things, there is no "better" or "worse" result, There is, though, an international standard that defines a "score" for the color error of a camera sensor in combination with a certain transform matrix). And in fact that standard even suggests a best transform matrix. But both of those are based on some wonks' decisions as to what spectrums best deserve to have their color correctly reported by the camera.

How do our cameras do on this? The score starts at 100 (for no error), and decreases for more error. As reported by DxO labs, for a Canon EOS 1D Mark IV, the score is 81.7 when working under D50 ambient (using the "best" transform matrix that is suggested by the international standard, given the measured behavior of the sensor itself). For a "tungsten" ambient, the score is 75.2.

So, some more food for thought.

Best regards,

Doug

Thanks Doug for the info! I figured something like that, but not in the terms you put it. More thinking on the de-mosaicing process and color interpretation by the app, but just that.

Leo

I don't get it though, the default in LR is IMO bad and I have my own defaults, ditto C1. You can't compare programs based on their defaults can you?

Ben Rubinstein said:

I don't get it though, the default in LR is IMO bad and I have my own defaults, ditto C1. You can't compare programs based on their defaults can you?

Click to expand...

Well... it's an start. If 2 programs interprets colors in different ways from the beginning, there are lot of probabilities that each one of them will return different colors when developing. For example, Lightroom tends to the reds (very noticeable on skin), and whatever you do to color correct it it always lefts some reddish tone on it. RPP, instead, gives natural skin colors from the beginning. I have been testing the newer versions of RAW Editors and they tend to interpret colors almost in the same way nowadays, they're all almost standardized, but RPP keeps the lead of doing a real color interpretation.

But RPP has a cons: for landscape/waterscape photography is not so good at colors like Lightroom or Aperture.

What Ditto C1 means? Is that a new app I'm missing?

Thanks for your comments!

Leo