5DmkII High ISO / no NR Workflow Comparison: LR vs CPP

Hello,

Long time lurker, first time poster here.

Since receiving my mkii body I've been wanting to to a test to compare how ACR and CPP stack up in converting raw files to jpg. Now that Adobe LightRoom 2.2 supports the 5D2 raw format, what better time for it

Test #1: High ISO work flow Comparison:
LightRoom vs Canon Photo Professional - No Noise Reduction Applied

-----------------------------------------------------------------------------------------------

Setting
----------

The camera was placed on tripod facing down onto a go board with a bowl of black go stones, about 3.5' away. Lighting was indirect daylight from a large-ish window on a cloudy day - very neutral.



Camera Settings
----------------------

• 5DMK2 body & Canon 50mm f/1.4
• RAW + JPG (Large, fine)
• Shots were taken at f/5.6 @ ISO 1600, ISO 3200 & ISO 6400
• Shots were taken using mirror lockup and 2 second delay on tripod.
• AWB
• image custom functions were at default:
• Long Exposure Noise Reduction - Auto
• High ISO Noise Reduction - Standard
• Highlight Tone Priority - Disable
• Auto Lighting Optimizer - Standard

Work Flow
---------------

Adobe LightRoom 2.2:

All CR2 files were imported to LR using the built-in converter. All noise reduction was removed. The files were then exported as a JPG (quality @ 90%) and TIF. Labels on the results image are as follows:

raw -> LR -> JPG
raw -> LR -> TIF

Canon Photo Professional 3.5.1:

All CR2 files were imported to CPP using the built-in converter. All noise reduction was removed. The files were then exported as a JPG (quality @ 90%) and TIF. Labels on the results image are as follows:

raw -> CPP -> JPG
raw -> CPP -> TIF

The files were then combined into one image which was then saved as a single TIF file.

Results:



To my eye, they are pretty darned close. Some slight differences I see:

• LR renders a touch more red-ish than CPP
• CPP either renders a little brighter, or has a little better dynamic range
• noise seems similar on both, CPP noise looks a little nicer, and will probably clean up a little easier.

Note that the top sample jpg is straight from the camera with default noise settings. In hindsight it would have been better to turn all the NR off in-camera before the test, but I didn't I don't normally shoot in JPG, so lets just say it's there for comparisons sake. Perhaps in my next test I will use it to compare post-processing noise reduction.

What do you think?

Thanks for the work, and for the careful controls.

Your results match for Photoshop what I found using the Canon raw files downloaded from Imaging Resource. I wish IR would provide more raw samples for the 5DII - they offer many more for the Nikon D3x, D700, etc. Currently, they have only the outdoor scene and one indoor test chart.

The results of the in-camera jpg show why so many people incorrectly thought the 5DII had great high ISO performance early on. The in-camera noise reduction is simply smashing half the detail. The 5DII's actual high ISO performance is a very small improvement over the 1DsIII, probably due to a thinner Bayer filter (which means more chroma noise downstream when you have to up the saturation, and really shows up at ISO 3200).

By the way, if you really want to see the differences between Photoshop and CPP, convert a few images with extremes of light, color and contrast. Night images of fireworks, Christmas trees, or building interiors will be particularly revealing. I use three different converters routinely (CS3, NX2 and DxO), because no one is best in all circumstances. I used to really dislike Nikon's Capture NX, but the user interface on CPP is even worse (in my opinion).

Again, thanks for the careful work.

Forgot to mention: camera profile "faithful" were used in both LR & CPP

Cheers,
Graham

Thank you Graham -

I both lack the time and energy to conduct such tests. I absolutely hate to work with DPP, but love to work with Lightroom and also have 2.2.

Before I got Lightroom 2.2, I shot everything RAW + Jpeg and processed only the jpegs, saving the RAW for the day that I might need it. I was amazed at the low noise. Now that I am processing RAW in Lightroom, I find more noise than in the jpegs.

http://wasillaalaskaby300.squarespace.com/

for a grand effort...

I guess the 5D2 is 'slightly' better than the 1DsMkIII ?

well and carefully done
IMO
=========

as more tests like your's come in...we can now .....either
return our cameras...

or go use them....and enjoy them

TOM

you're welcome

I wanted to do this test to see if there was a discernible difference between using LR (my preferred method) vs Canon's software. I had read some posts stating that LR was much worse in it's conversion of the raw files. With what I've seen today with this test, I won't hesitate to use LR in the future. The differences are negligible.

Cheers,
Graham

Adobes raw conversion noise reduction is no up to par yet. DPP is such a pain to use but until adobe and the other raw conversion programs get their act together on the 5d mkII I'm going to be stuck with it.

Columbusshooter said:

Adobes raw conversion noise reduction is no up to par yet. DPP is
such a pain to use but until adobe and the other raw conversion
programs get their act together on the 5d mkII I'm going to be stuck
with it.

Click to expand...

Agreed - DPP is clunky.

Any reason you don't use NeatImage, Noise Ninja or another NR tool?

the Canon conversion software?

-evan

--
I do know how to spell. I'm just a lousy Tipyst!

http://www.pbase.com/eheffa

Evan Effa said:

the Canon conversion software?

-evan

Click to expand...

Yep, CPP should read DPP. Sorry for the confusion.

Graham, others:

I think you will find that some third party noise busters are much, much better than what comes with Adobe, Nikon or Canon. I own and use NeatImage, NoiseNinja and Define. Of the three, NeatImage gives me the best results most of the time. It provides a very nice combination of area noise reduction with sharpening, and produces very few artifacts. I find that a very light pass of NI can knock about one full EV off noise, turning an ISO 1600 image into ISO 800. For the money, it's cheaper than buying a newer camera.