Is RAW much better?

I was just confirming my observation. I could never wrap my head around whether that was a good thing or a bad thing. I know I use SRGB monitors. Printer profiles? Windows vs Mac color profile implications?? So I just shoot SRGB. Not because I understand why. Your photos always look fantastic to me.

Zee Char said:

dprsok said:

So you are using ADOBE RGB?

--
http://www.flickr.com/photos/39169343@N04/

Click to expand...

--
The solution is always simple. Getting there is the hard part.

Click to expand...

--
http://www.flickr.com/photos/39169343@N04/

For best results -

Shoot RAW, capture to aRGB for the widest gamut on capture, edit in aRGB or ProPhoto for a wide gamut edit space. Print in aRGB. Convert to sRGB for the web.

Modern printers such as the Canon Pro9500II can print more than your monitor can display, nearly the entire aRGB gamut.

dprsok said:

I was just confirming my observation. I could never wrap my head around whether that was a good thing or a bad thing. I know I use SRGB monitors. Printer profiles? Windows vs Mac color profile implications?? So I just shoot SRGB. Not because I understand why. Your photos always look fantastic to me.

Click to expand...

Peter 13 said:

Luke Kaven said:

[...] and you're acting as though the least significant 6 bits don't represent full stops of exposure just the same.

Click to expand...

We are talking about 16 bit vs. 8 bit TIFFs now? Exactly. They do not. They represent more intermediate values. Don't you have PS on your computer to play with?

Click to expand...

At some point, you can remap bits to code for any value you want. But the sensor is linear. And a linear capture or approximately linear, depending upon the software. I use C1 (and PS of course). I no longer have a 5DII and now use a D3x.

Peter 13 said:

Peter 13 said:

Peter 13 said:

Peter 13 said:

As far as being "highly" non-linear, I think that is somewhat exaggerated. The addition of a tone curve is not that significant here.

Click to expand...

Let us see. I was talking about the curve during RAW conversion. Here is a well exposed pic saved as a linear TIFF:

Click to expand...

If this is a linear capture with no tone curve explicit or implied, then it is underexposed. The linear capture is the accurate capture . So if your linear capture is dark, you underexposed.

Click to expand...

I see. You have never seen how a linear TIFF, displayed 1-1 on your sreen, looks like. Open DPP, take any well exposed RAW, and check the "linear" box.

Click to expand...

The sensor is linear. The only accurate numbers are the linear ones. If you aren't actually exposing to the right for the sensor values themselves, then what else? The linear data is the pristine source, the archival data, the clean capture. Have a look at the UniWB concepts and let us know what you think.

Luke Kaven said:

Peter 13 said:

Luke Kaven said:

[...] and you're acting as though the least significant 6 bits don't represent full stops of exposure just the same.

Click to expand...

We are talking about 16 bit vs. 8 bit TIFFs now? Exactly. They do not. They represent more intermediate values. Don't you have PS on your computer to play with?

Click to expand...

At some point, you can remap bits to code for any value you want.

Click to expand...

Bits are just bits. Glad that we agree on that. But there is a convention how the 16 bits are used. They are used for finer steps, not for a greater DR.

Luke Kaven said:

But the sensor is linear. And a linear capture or approximately linear, depending upon the software. I use C1 (and PS of course). I no longer have a 5DII and now use a D3x.

Click to expand...

The sensor plays no role in this argument. You can have a picture created by software, for example.

Luke Kaven said:

Luke Kaven said:

I see. You have never seen how a linear TIFF, displayed 1-1 on your sreen, looks like. Open DPP, take any well exposed RAW, and check the "linear" box.

Click to expand...

The sensor is linear. The only accurate numbers are the linear ones. If you aren't actually exposing to the right for the sensor values themselves, then what else? The linear data is the pristine source, the archival data, the clean capture. Have a look at the UniWB concepts and let us know what you think.

Click to expand...

Only after you tell me what you think about the effect of clicking on "linear" in DPP.

Luke Kaven said:

Shoot RAW, capture to aRGB for the widest gamut on capture, edit in aRGB or ProPhoto for a wide gamut edit space. Print in aRGB. Convert to sRGB for the web.

Modern printers such as the Canon Pro9500II can print more than your monitor can display, nearly the entire aRGB gamut.

dprsok said:

I was just confirming my observation. I could never wrap my head around whether that was a good thing or a bad thing. I know I use SRGB monitors. Printer profiles? Windows vs Mac color profile implications?? So I just shoot SRGB. Not because I understand why. Your photos always look fantastic to me.

Click to expand...

Click to expand...

I was just going to explain that. There were some interesting threads on this about 5 years ago. Thats what the Digital Dog (Andrew Rodney) and other posters recommended so I started doing that. I purchased his book.

http://digitaldog.net/

--
The solution is always simple. Getting there is the hard part.

dprsok said:

I was just confirming my observation. I could never wrap my head around whether that was a good thing or a bad thing. I know I use SRGB monitors. Printer profiles? Windows vs Mac color profile implications?? So I just shoot SRGB. Not because I understand why. Your photos always look fantastic to me.

Zee Char said:

dprsok said:

So you are using ADOBE RGB?

--
http://www.flickr.com/photos/39169343@N04/

Click to expand...

--
The solution is always simple. Getting there is the hard part.

Click to expand...

--
http://www.flickr.com/photos/39169343@N04/

Click to expand...

Nothing wrong with sRGB.

--
The solution is always simple. Getting there is the hard part.

bellevilleguy said:

To answer the OP's other question about converting to .tiff from RAW, I always use DPP to do this, in my opinion nobody knows a RAW file like Canon's own software, so I use it to do the converting... I use PhotoShop to do the rest of the work to the .tiff, before I once again convert it into .jpg, again using DPP to do this... This is my recipe, and what works for me!!

Click to expand...

It's possible that this was once true, but I think the noise reduction in the version of Adobe Camera Raw that shipped with CS5, LR3 and (I think) PSE8 is much better than DPP now.

All I can say is if Canon knows their raw files better than anyone else, then why is their jpg conversion so horrendous? I can take an ISO 6400 raw file out of my 6400 and with a little noise reduction in ACR, it's gorgeous. However, the jpg version of it is just riddled unusably with noise.

1) You argue that bits are not stops. I reply that in theory you are right.

2) I point out that in practice, since bayer sensors are linear, the numbers that come out of the sensor A-D converters are linear. Therefore in terms of RAW image data, the bits are stops (assuming no preprocessing of RAW data).

3) You suggest to me that the linear mapping is underexposed. You appear to be judging it with respect to various tone curves (portrait, landscape, etc) which produce a brighter appearance.

4) I remind you that the linear mapping is the first order function, and the tone curves are the second order functions. The linear numbers are the true indication of the level of pixel saturation and are reliable indicators of exposure. You should judge your exposures based upon the "linear" or "neutral" histogram. If you expose to the right, highlights in the linear rendering will not appear dim.

5) As far as everything I know, using C1 (and I'd think the others) to capture to an 8 bit TIF gives you the high 8 bits after capture and truncates the rest. In the trivial sense, this is "scaling" (by bit shifting), but the intermediate values will always be padded zeros. But it seems you are saying something other than this. I'm not clear on what else you think is the case.

Luke Kaven said:

I apologize for the tone I took earlier. I do feel you may have misled people with your advice however.

You asked about my images; they're available at a link under my profile.

Click to expand...

No problem Luke.

I do think you missed my quotes at the end of my posts tho.

1:This is just my opinion. People who think differently have every right to think the way they do, just as I have.
and
2:My advice is that you use what you are happy with
As you can see I never advised people to use JPG, just told them what I use.

Must go and checkout your images.
Very best regards,
--
Sel ................
To me, PC means personal computer, not politically correct.
http://flickr.com/photos/selsphotos/
http://sel.photosales.co.nz

Luke Kaven said:

3) You suggest to me that the linear mapping is underexposed. You appear to be judging it with respect to various tone curves (portrait, landscape, etc) which produce a brighter appearance.

Click to expand...

No. Take a look here:

http://www.adobe.com/digitalimag/pdfs/linear_gamma.pdf

Luke Kaven said:

5) As far as everything I know, using C1 (and I'd think the others) to capture to an 8 bit TIF gives you the high 8 bits after capture and truncates the rest. In the trivial sense, this is "scaling" (by bit shifting), but the intermediate values will always be padded zeros. But it seems you are saying something other than this. I'm not clear on what else you think is the case.

Click to expand...

Each RAW conveter applies a very non-linear curve that clips part of the DR. Let us say that the calculations are done in 16 bits, and saved as such. If you choose 8 bit, then the steps are bigger but the DR is the same. If you play with PS, you will see that it is basically a division by 2^8. You think of this as a shift of the DR. It is not, because the viewers know that the new file is 8 bit, and know how to interpret the new nunbers. If you can trick the viewer into thinking that the new 8 bit file is 16 bit, then yes, you will get a lower DR.

Another experiment yo can do - save an 8 bit TIFF as 16 bit, without dithering. Then apply some non-trivial curve to it and to the original 16 bit. You will see the same DR but gaps in the converted file.

For many people who came up shooting print and slide film, it is natural to think of shooting with JPEGs. This is of course the venerable craft of photography. You take the picture, and you get the results after "developing" the image. This is how I came up too, and we're approximately the same age.

But the business of shooting RAW is more than "the thing to do" these days. I just want to make sure that others who are just starting out today do not limit their future potential by throwing away their archival data in favor of instant pictures -- not before they realize the value of what they are throwing away.

My most recent "you got the job" moment came when I was shooting a portrait of someone in front of a stained-glass window. I had to expose to not blow out the window, but the subject was in silhouette. There was plenty of room in the 14-bit RAW file to bring up the subject -- cleanly -- in the face without noise. In the JPG, it would have been lost. As I said, this was a "you got the job" moment. If I seem sensitive to what advice people get, it's because I want them to achieve their potential.

In the Adobe doc about linear gamma, much of what I'm saying is under the section labeled "Linear Capture". Everything I've said is compatible with that.

You should use linear (or neutral) settings in camera so that the preview JPEG accurately reflects the exposure. Again, the sensor and A-D are all linear to a best approximation. The first-order data is linear, almost as good as a photon count. Once you begin looking at anything with a tone curve of any kind, you are looking at a second order function, not a true indication.

Is any part of this in dispute?

If you are exposing to the right for the linear data, you will get maximum S/N and maximum quality. The linear capture will not appear dim, just linear. This is a psychological barrier. All this is compatible with Adobe's article. And all this is detailed in the UniWB description.

Luke Kaven said:

In the Adobe doc about linear gamma, much of what I'm saying is under the section labeled "Linear Capture". Everything I've said is compatible with that.

Click to expand...

I have never disputed that because (1) it is trivial and (2) has nothing to do with 8 vs. 16 bit TIFFs.

Luke Kaven said:

You should use linear (or neutral) settings in camera so that the preview JPEG accurately reflects the exposure.

Click to expand...

There are no such settings in any camera that I know of. The "Neutral" setting is just a marketing term, the heavy non-linear curve has been applied already. You can see the curve on p.2 of the article. If you replace it with a linear one, you get the flower image on p.1.

Really, what part of this do not you understand?

Luke Kaven said:

Again, the sensor and A-D are all linear to a best approximation. The first-order data is linear, almost as good as a photon count. Once you begin looking at anything with a tone curve of any kind, you are looking at a second order function, not a true indication.

Click to expand...

"Not a true indication" are the pics that we all view and enjoy.

Peter 13 said:

Zee Char said:

http://www.filedropper.com/mg9559_1

It did. Well everyone give it your best shot. I'll be looking to seeing the results when I get back.

Click to expand...

Here are my conversions. I tried to open the shadows a bit. I applied different amounts of NR.

http://www.filedropper.com/pics

Click to expand...

Thanks to dprsok and Peter 13 for working on and submitting images. People told me I did not know what I am doing but I'm not yet convinced that I need to spend $200 + on light LR as I just upgraded to CS5. I was hoping to see more people submit some conversions to help me spend some money

--
The solution is always simple. Getting there is the hard part.

Zee Char said:

Thanks to dprsok and Peter 13 for working on and submitting images. People told me I did not know what I am doing but I'm not yet convinced that I need to spend $200 + on light LR as I just upgraded to CS5. I was hoping to see more people submit some conversions to help me spend some money

Click to expand...

CS5 does the same job with RAW conversion as LR3, AFAIK. I do not use the catalog features of LR, so IMO, you do not need LR.

RedFox88 said:

Zee Char said:

Well I did say that but further on I also said it was important to me for my wildlife images - particularly crops.

Click to expand...

RedFox88 said:

One from the zoo the other day.

Click to expand...

Since when are zoo animals "wild"-life?!

Click to expand...

OK Red. I got kinda bored tonight. Here you go. Some non zoo critters.





































--
The solution is always simple. Getting there is the hard part.

Peter 13 said:

Zee Char said:

Thanks to dprsok and Peter 13 for working on and submitting images. People told me I did not know what I am doing but I'm not yet convinced that I need to spend $200 + on light LR as I just upgraded to CS5. I was hoping to see more people submit some conversions to help me spend some money

Click to expand...

CS5 does the same job with RAW conversion as LR3, AFAIK. I do not use the catalog features of LR, so IMO, you do not need LR.

Click to expand...

Yes I should have specified ACR. There are a lot of people who use and recommend LR because of all the features it has so that just came out. Thanks for the correction.

I'm not sure I mentioned this so forgive if I have. I think there was a Canon survey through DPreview a few years ago. May have been somewhere else. There was a section on what improvements you would like to see and one from me was "If DPP came out with their engine and all the LR features I would pay big bucks for it".

--
The solution is always simple. Getting there is the hard part.

Zee Char said:

Peter 13 said:

Zee Char said:

http://www.filedropper.com/mg9559_1

It did. Well everyone give it your best shot. I'll be looking to seeing the results when I get back.

Click to expand...

Here are my conversions. I tried to open the shadows a bit. I applied different amounts of NR.

http://www.filedropper.com/pics

Click to expand...

Thanks to dprsok and Peter 13 for working on and submitting images. People told me I did not know what I am doing but I'm not yet convinced that I need to spend $200 + on light LR as I just upgraded to CS5. I was hoping to see more people submit some conversions to help me spend some money

Click to expand...

I bought LR3 and CS5. I like LR3 a lot. But CS5? The cryptic methods are a bit much. I get the feeling that one really needs formal training to use it effectively. I know that pros that have used it for years don't have any issues with the concepts. But the concepts are ancient. For all the tutorial videos that explain how to accomplish a certain task, I haven't found any that attempt to explain:
What a layer is
What a mask is
How are they different
How do they interact with each other, and so on.

I like LR3 with its concept of all adjustments being in a sidecar. The virtual copy simply creates an additional sidecar. I found that there are certain adjustments that I use on every photo. So I created a preset containing all those adjustments (including Lens Correction). I can apply the preset to one photo, or all selected photos. Once applied, I adjust Luminance NR and Sharpness from my preset defaults, address blown highlights and lowlights, tweak Exposure, Fill Light, possibly add ND Filter, Crop / Straighten and maybe use the adjustment brush. Conversion to JPG also will add output sharpening.

I know that eventually I will find myself transferring to CS5 from LR3 to take advantage of CS5's more powerful features. So far I have used the content aware fill for removal of distracting objects. It's too bad that CS5's adjustments can't be sent back to LR3 as part of the sidecar. After using CS5, I end up with my raw file (CR2), the LR3 sidecar containing the LR3 adjustments, and a TIFF created from the CR2 with LR3 adjustments and CS5 adjustments.

--
http://www.flickr.com/photos/39169343@N04/

If if it is an acronym what does it stand for ? If it is a word why don't you write "raw"?

It is a type of file. But not a file type.

--
http://www.flickr.com/photos/39169343@N04/