Rainy Saturday; Raw Study

[aparmley]

After returing from the DMV [God I love that nut house] I decided to focus on RAW today, expanding my understanding and editing abilities. It is a rainy day and seeing that I just picked up a book on RAW I figured what the hey.

So here are some samples along with textual comments. I have learned today why I tend to have more noise in my RAW images than JPEG - I wanted to do my noise reduction either in PS or in neat image, but Adobe RAW has noise reduction tools, I just wasn't familiar with them and always elected to not use them. I used them today and I found a solution to my noise problems and thought I would share.

First off lets look at an image shot at ISO 100:

http://parmley.smugmug.com/photos/66888130-L.jpg

My edits for the above image were like so:

Raw conversion tool edits:

Whibal WB
-.30 EC
Brightness 60
Contrast +15
Saturation +5


Photoshop edits:

Levels
USM 75 | 1 | 1


Here is the same image taken at ISO 1600 with only Whibal WB set; no other edits:

http://parmley.smugmug.com/photos/66877617-L.jpg

The following image is an edit of the above ISO 1600 image:

http://parmley.smugmug.com/photos/66880616-L.jpg


My edits for the above ISO 1600 image were like so:

Raw conversion tool edits:

Whibal WB
-.30 EC
Brightness 60
Contrast +15
Saturation +5
Luminance noise reduction 25/100 [setting are 1 to 100; this indicates I used a setting of 25]
Color Noise Reduction 10/100


And finally I also did a few additional edits to the above image in photoshop and it resulted in this image:

http://parmley.smugmug.com/photos/66887494-L.jpg

My edits for the above ISO 1600 image were like so:

Raw conversion tool edits:

Whibal WB
-.30 EC
Brightness 60
Contrast +15
Saturation +5
Luminance noise reduction 25/100 [setting are 1 to 100; this indicates I used a setting of 25]
Color Noise Reduction 10/100


Photoshop edits:

Levels
USM 75 | 1 | 1


so if we look at the ISO 100 image next to the ISO 1600 images before noise reduction and the ISO 1600 after noise reduction we can see their differences more magnified in the following image:

http://parmley.smugmug.com/photos/66893428-L.jpg

I apologize I don't know of a good way to maintain the exact same size of a high magnification crop, I try to eye it but its never the same size. . . . But hopefully its large enough for you too see that images on the far right, the ISO 1600 with noise reduction cleaned up better - but its still no ISO 100 shot, but compared to the straigh ISO 1600 shot, its a lot smoother.

So I learned how to manage noise better when shooting RAW. I'm happy with this accomplishment today. Luminance noise seems to be the most dominant noise in my images and it is the kind of noise that adds grain - so I on paper - my image from here on in should be more noise/grain free now that I can eliminate it from a 16bit RAW image before I even open it in Photoshop. . . Will see if that holds true but I happy that I found the solution to my noise issues with RAW. . . Maybe this will help others out.

[aparmley]

XT | 50 1.4 @ 2.0 | ISO 1600 | 1/30 [Raw Conversion]
http://parmley.smugmug.com/photos/66914651-M.jpg

XT | 50 1.4 @ 2.0 | ISO 1600 | 1/40 [Raw conversion]
http://parmley.smugmug.com/photos/66914633-M.jpg

[ReF]

interesting.

i think i mentioned this before but sometimes i find more noise when converting RAW images using adobe's converter over canon's. i also went through a phase where i was using adobe RAW's noise reduction feature for very noisy pictures - then when i started using noise ninja i had to go back and re-edit all of those photos because the results that much better. noise ninja was on a completely different level IMO.

[arigato]

I belive raw images are 12 bit (not 16) vs 8 and are better at capturing a wider dynamic range because of it if I remember correctly according to an older post. I think Coldrain could probably explain that better.

[timmciglobal]

Yea, raw have 1 stop advantage in DR and exposure recovery. You'd be amazed try overexposing by 2 stops then bringing the picture back down in ACR. Try same thing in jpeg, much harder and more blown highlights depending on the scene.

Tim

[aparmley]

Adobe RAW opens Canon CR2 files as 16-bit files - Or I could override that and have it open at 8-bit, I don't know where the 12-bit number you mentioned comes from - maybe I am missing something here?

In all actuality RAW does not give you more DR. to qoute bruce fraser, "Dynamic range is an analog limitation of the sensor." However, i think some people mistake the increased bit depth RAW gives you as an increase in DR, which isn't so. The two are related, indirectly. Bruce fraser uses the staircase analogy, "imagine a staircase, the DR is the height of the staircase. THe bit depth is the number of steps in the staircase. . . more bits, or a larger number of smaller steps, does not increase DR, or the height of the staircase."

Yea, raw have 1 stop . . . and exposure recovery

Do to the linear capture of CCD/CMOS sensors you can get more exposure recovery if you over expose RAW. The sensor using half of its available photons to record the brightest stop of light, then half of the remaining half for the next stop darker and so and and so forth until the darkest shadow of the image only contains something like 64 out of the available 4,096 photons. I never knew about the term linear capture until reading bruce frasers book - it was very interesting. Now , I understand why you're able to recover over exposed images better, because more detail ends up in the midtones and shadows than if you were to try to recover and underexposed image, less detail ends up in the midtones and highlights. Thats my understanding of it all anyway.

ReF - I've stopped using neat image - it messes with my colors.

Top image Photoshop reduced noise; bottom neatimage
http://parmley.smugmug.com/photos/67057898-L.jpg
http://parmley.smugmug.com/photos/67057890-L.jpg

As you can see the colors are little flat with the neatimage shot. if I absolutely want a clean(er) image I'll use neat image, but if colors are more important to me I won't.

I D/Led noise ninja but their trial offer puts text all over the photo - I abhore that. so I won't pony up for the home version just so I can see what my images will look and print like with their product. :mad:

[coldrain]

He meant that the sensor only gives a 12bit value per colour (R, G, B). That Photoshop puts that into 16 bits per colour only has a reason in memory management. It is much faster calculating and moving 16 x (3+1) = 64 bits per pixel than it is to move 12 x (3 + 1) = 48 or 14 x (3 + 1) = 56 bits per pixel.

The extra bits next to the 3 colours are for transparancy, also called "alpha channel", something Apple came up with long ago.

So... RAW delivers 12 to 14 bits per pixel on the moment (depending on the sensor) and only one colour per pixel, the rest of the colours get interpolated by the RAW conversion (this is why RAW convertors can give such different results from eachother). And that gets stored after the other colour values are determined from the neighbouring pixels in 64 bits, where 16 or 8 bits of those (depending on the sensor) contain no information until you start to manipulate the photo.

[aparmley]

He meant that the sensor only gives a 12bit value per colour (R, G, B). That Photoshop puts that into 16 bits per colour anly has a reason in memory management. It is much faster calculating and moving 16 x (3+1) = 64 bits per pixel than it is to move 12 x (3 + 1) = 48 or 14 x (3 + 1) = 56 bits per pixel.

The extra bits next to the 3 colours are for transparancy, also called "alpha channel", something Apple came up with long ago.

So... RAW delivers 12 to 14 bits per pixel on the moment (depending on the sensor and only one colour per pixel, the rest of the colours get interpolated bythe RAW conversion, this is why RAW convertors can give such different results from eachother). And that gets stored after the other colour values are determined from the neighbouring pixels in 64 bits, where 16 or 8 bits of those (depending on the sensor) contain no information until you start to manipulate the photo.

Thanks for the explanation. . . ;)