Pages: [1] 2 3 4   Go Down

Author Topic: Quick Dynamic range comparison - Leaf AFi-ii 12, CFii-39MS, Phase P20, Canon 5D2  (Read 18925 times)

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

I had all the cameras sitting here and wanted to make a quick test to see how they performed for my own curiosity and decided to share the results here.  I am only presenting images of the step wedges and do not have any quantitative data to share because I don't have a Mac version of Imatest yet (Hey Norman - how about that beta?) and wasn't at the studio to use the light table so this was just a quick test using my NEC monitor to light the step wedge - definitely not perfect.  The idea was to shoot the stoufer transmission step wedge and see how much the shadows could be brought up.  I exposed to get the #1 wedge as close to clipping as possible when the file was set to linear with no other adjustments.  I then imported all the files to Lightroom and reset curves in each file to linear, blacks and both color and luminance noise reduction all to zero.   I am sure that were I to use C1 for the phase files or phocus for the CF files I might get better results so this is just for a quick comparison.  Each step in the stouffer transmission chart is about 1/3 a stop, but more importantly how does each file hold up and how much can the files be lifted?

« Last Edit: May 19, 2011, 02:11:43 am by EricWHiss »
Logged
Rolleiflex USA

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

I was a bit surprised how close they all looked  - except for the P20 file - it looked correctly exposed in C1 but is showing down a stop in LR3 for some reason.  Anyhow, examine next the same files pushed up 3 stops in exposure in LR3.  This is where you can see the real differences. I could have pushed the CFii-39 and AFi-ii 12  up to 4 stops but the canon file already looked terrible.

« Last Edit: May 19, 2011, 02:00:34 am by EricWHiss »
Logged
Rolleiflex USA

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

Just for fun, I added a set with the CF-ii 39 back set to multishot mode.    btw - The spots outside the step chart are really marks in the black matte board.
Logged
Rolleiflex USA

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

Notes:  
The CFii-39 in multishot mode seems to be the best overall.  
The Leaf file is really large but the noise profile is pretty good zoomed in and is probably the best single shot. At 4 stops exposure adjustment the two halves of the sensor are visible.
The Leaf file is hard to white balance because it's changing from light to dark.  Balance the brightest side and get a red shadow tone, or balance shadows and get a green tone (regardless of input color profile).
The Canon 5D2 has obvious pattern noise and not usable pushed 3 stops
The lens used for all the digital backs was the Rollei 90mm apo macro.  
I do not plan to make the RAWs available, but may provide 100% crops at a later time - but not sure they are needed.

I would like to see the IQ 180 at the lower ISO 32 setting and also with sensor plus activated at base ISO.
« Last Edit: May 19, 2011, 01:58:02 am by EricWHiss »
Logged
Rolleiflex USA

Peter Devos

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 518

I really love my Hassie 39Mp. There is always a lot of people saying Phase and Leaf are better than Hasselblad, but as you Eric, i always try myself and came to the conclusion that Hasselblad is no way worse than Phase or Leaf. That is why so many professionals are using Hasselblad. Same with the lenses. I have had the three schneider LS on a DF...... no better than the Hasselblad HC lenses. ::)
Logged

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

Yes, I agree with you Peter.   ;)    
Actually, I think the best back I've used is still the ixpress 528c (at least at base ISO). I really liked the color and DR.  Now that I've had both the 528c and the AFi-ii 12, I can be sure the 528c did outperform the Aptus 12 in microstep mode.   I'm sort of experimenting with lenses a bit now - haven't tried any of the H lenses yet but did test the Phase (and Schneider) 80mm LC and thought it was okay - but just that.
« Last Edit: May 19, 2011, 02:51:49 am by EricWHiss »
Logged
Rolleiflex USA

Peter Devos

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 518

I really love my 528c also but i will be selling it. I just have to make some pictures of it. I also use my 384 very often because i really love the square format. I hope to get a 5000 euro for my 528c with hasselbladH or HasselbladV adapter. The microstepmode is not surpassed by any back i have ever seen.
« Last Edit: May 19, 2011, 06:07:33 am by Peter Devos »
Logged

dergiman

  • Full Member
  • ***
  • Offline Offline
  • Posts: 206
    • http://www.philippderganz.com

A lot of the information in the files from the Canon 5d2 is in the highlights whereas my digital back doesn´t respond to highlight recovery half as good. There is not much to gain by pushing a file from the 5d2 by 2 stops or more.
Logged
Philipp Derganz Photography
www.philippderganz.com
Architektur Photo
Wien, Austria

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

A lot of the information in the files from the Canon 5d2 is in the highlights whereas my digital back doesn´t respond to highlight recovery half as good. There is not much to gain by pushing a file from the 5d2 by 2 stops or more.

The exposure in the test is set such that the top wedge is the highest possible without clipping *in linear mode* therefore there is nothing to recover.  The exercise shows the total range possible - either its there or it isn't.  If I set the exposure higher, the first few steps look the same and don't recover.   But of course you are absolutely free to set up your own tests if you think there is something more here to discover.
« Last Edit: May 19, 2011, 11:14:19 am by EricWHiss »
Logged
Rolleiflex USA

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051

Thank you for doing this test. How do you suggest I should interpret the (+3 stops) image files? Is the thing to look for when one no longer can discern discrete steps towards black? Or when the noise level become unbearable?

It would have been interesting to see how the Nikon D7000 would do in this test.

Can a monitor be "trusted" in terms of flickering etc for this kind of measurement?

Since this test is artificial, how relevant do you think it is for more realistic scenes (where, if any, in-camera noise reduction would have a harder time)?

Perhaps an improvement to this test would have been analyzing the linearity from black to white (spatial averaging), and the noise level. A perfect camera (in this sense at least) then would have an even staircase response from 1 to 41, and no noise (variance) within.

-h
Logged

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

That's a good question and you're welcome.  In looking at the files, I just try to imagine my own images with the shadows lifted that much and what noise I would see in the images.  Originally I intended to post images with the shadows lifted or fill light setting set to max, but this method differs so much from one application to the next that I used exposure instead.  I also took some real life comparison shots which I may post soon. This will help show the same thing.

ps. I forgot to add exposure times were over a second so flicker was not a concern.
« Last Edit: May 19, 2011, 12:10:51 pm by EricWHiss »
Logged
Rolleiflex USA

Gigi

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 548
    • some work

Eric -

Seems like there are three parts to this:

1) the making of disciplined real-world apples to apples comparisons from one back to another
2) the technique used to make the test - is it good enough? Is the lighting a factor?
3) reading the test - what do we look for?

To read the test, it seemed we should look at the "edge lines" from step to step, and when they were lost.  I'm not quite understanding the lighting technique behind the test yet.

But boy, its great to see actual hands-on "torture tests" from back to back. We rarely see that, and thanks for doing this!

Geoff
« Last Edit: May 19, 2011, 02:40:12 pm by Geoffreyg »
Logged
Geoff

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

Hi Geoff,
Yes that's more or less it - look to see where the dark steps become distinct but also look at the noise level and see if that is usable or not.  In a real image, one might have a reason to lift shadows or adjust exposure and this is just an easy way to compare as the lighting was constant and the wedge gives us a usable measure of the DR.  eg. 30 steps is 10 stops.   
Logged
Rolleiflex USA

Doug Peterson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4210
    • http://www.doug-peterson.com

I am sure that were I to use C1 for the phase files or phocus for the CF files I might get better results so this is just for a quick comparison.  

Eric, what a fun test (at least for us nerds).

Two thoughts:
- I would have loved to see a textured material or other detailed subject matter alongside the smooth stepper chart. Obviously harder to find a stepper pattern with texture/detail, especially the 3 dimensional (rather than printed) kind.
- I would love to compare the Canon, Phase, and Leaf files in Capture One vs. LightRoom vs. Etc (my interest dwains quickly after that, but Aperture, Irrident Raw Developer would be my next two). In particular the pattern noise in the Canon, in my experience, is much better handled in Capture One.

Part of the real world application of dynamic range, in my opinion, is not only how much noise you get, but what KIND of noise and how well it can be handled by the raw processor of your choice, and how that noise+raw_processor affects the overall look of the image, tonal transitions, and detail. That's why I say I'd love to see both textured/detailed material as well as including software in the mix, especially when Hassy and Team Phase One (Leaf/Phase) but a lot of effort into eaking out that last bit of detail/information in the file, often through proprietary/dedicated means (e.g. the black cal file).

I'd much rather have those two things than "quantitative" Imatest data which is quite one dimensional in it's real world application.

Still, I want to emphasize my appreciation for your test and your efforts. I know all too well that no matter what test you run or how carefully/well-meaning someone (actually many people) will find something they would have done differently.

Doug Peterson (e-mail Me)
__________________

Head of Technical Services, Capture Integration
Phase One Partner of the Year
Leaf, Leica, Cambo, Arca Swiss, Canon, Apple, Profoto, Broncolor, Eizo & More

National: 877.217.9870  |  Cell: 740.707.2183
Newsletter | RSS Feed
Buy Capture One 6 at 10% off
« Last Edit: May 19, 2011, 12:32:30 pm by dougpetersonci »
Logged

BobDavid

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3307

No wonder many of the finest museums use the CF39 MS. Put an APO Digitar in front of the back and the results are nothing short of breathtaking.
Logged

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

Doug,
Thanks and as you can imagine I did look at the files in C1 because that is what I mostly use for everything except the hasselblad files which aren't supported.  I looked at NR and shadow lifting for each file in several applications and there are big differences of course and though I'd like to show all of that it's just too much work.  I think C1 does a great job with the 5d files overall but LR3 does seem to really have great NR.   That's certainly important to figure into these tests because in a real image you would use some NR and I've presented here only images with NR off.

I'll post some real images and crops of the shadow areas sometime in the next days.
Logged
Rolleiflex USA

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

Here are some examples of how this shadow lifting would look with real images.   I took the same scene from my window yesterday with the 4 different backs.  I tried to expose the images as close to but not blowing out the white plastic bag in the center of the frame.

First the Hasselblad CF-ii 39MS back with 60mm Schneider Curtagon lens:
First shown whole frame images with no adjustments, and shadows lifted to 100%
« Last Edit: May 19, 2011, 04:20:40 pm by EricWHiss »
Logged
Rolleiflex USA

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

Here are some zoomed crops of the darker areas after being lifted with the shadows tool, note that I did not change any NR or other settings - only the shadows.
« Last Edit: May 19, 2011, 04:27:35 pm by EricWHiss »
Logged
Rolleiflex USA

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

Here is the same set of images in the same format for the Canon 5D2.  Shot with the 24-105L lens, and only adjustments were to raise the shadows slider to 100.  No NR.    Note that its still pretty dark.  I think this is where you can see that MFDB's have more DR.  A portion of the white plastic bag is blown out in this image already so I could not expose more or I would lose information on the top.  All of the digital backs have more room in the shadows without blowing out the bag.
« Last Edit: May 19, 2011, 04:52:17 pm by EricWHiss »
Logged
Rolleiflex USA

EricWHiss

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2639
    • Rolleiflex USA

And lastly from the Leaf AFi-ii 12

Logged
Rolleiflex USA
Pages: [1] 2 3 4   Go Up