Test Results: Pentax 67 vs Canon 5D

I photographed the same scene with each camera. The details and results are posted here: americansouthwestphotos.com/comparison
Marty Hulsebos

 

Many tests have confirmed that a good 10-12mp camera is equivalent to 645. But since 6x7 has twice the area of a 645 it would take the new 21mp 1Ds Mark III to equal 6x7. The Pentax is cheaper.

 

Thanks for the test. The Pentax does look more detailed. The Canon appropriately sharpened still holds up pretty well at that size and I would think this would be adequate resolution for quite large prints.

It's like the example on another thread here of a flatbed vs a Nikon MF scanner- sharpen the heck out of the former and it's hard to tell the difference between the two.

 

"Clearly the Pentax is the winner."

I've been using a Pentax 67II for about 6 years, and it's a fantastic camera. I've been using a 5D for about a year. Anyone who is interested in buying my P67II email me.

 

I had to upsample the 5D image

I used Photoshop's bicubic smoother

went through Camera Raw, with no sharpening or noise reduction

...how could this test could be reliable?

Anyone know that;

1_Any digital image need a bit of capture sharpening to get the perfect capture result.

2_That when you upsample a digital image it become blurrier, therefore you also need at that stage to apply a bit of sharpening to regain some details

Send me the file and i will redevelop it appropriatly so whe can compare apple vs apple correctly.

Im not implying that a 5D is better than a Pentax, i say that if you dont know how to develop & prepare a digital file the test result coudltn be accurate, like in your test. Maybe with better care the difference will not be that bad..then maybe you will like even more the faster way to shoot in digital ; )

 

Although I continue to love my Mamiya, I will point out that Nikonscan software applies sharpening, EVEN WHEN IT IS TURNED OFF. Also, all digital fles, scans or DSLR images, require sharpening.

 

Sharpen that 5D image and it is suprisingly close to the Pentax 67.

 

Have a look at this comparison of Canon 1Ds and Pentax 6x7 on Luminous Landscape, from a few years back:

http://www.luminous-landscape.com/reviews/shootout.shtml

 

A problem with this test is that both images were taken at f/8, even though the lenses were different focal lengths (105mm and 50mm). The depth of field will therefore be different. Unless the point examined was perfectly in focus, the Pentax 67 image will be softened (more than the 5D image) by lack of focus.

I'd suggest doing it again with the 50mm at f/5.6 and the 105mm at f/11.
That way the depth-of-field would be the same.

 

To respond to those who asked to see a sharpened version of the 5D, I have uploaded it to the site, along with a sharpened version of the Pentax 67 image. Here's the link.
Marty Hulsebos

 

I agree with Radford - you should reshoot with each camera focused on the same point and
different apertures to capture the same DOF.

 

I think there is plenty of depth of field for both images so that the crop area is well within the
range of clear focus. In fact it is probably very close to the actual focusing distance.

Marty Hulsebos

 

Whether due to focus/DOF or other problems, your images from both the 5D and the 67 seem rather soft to me. Possibly your focus is off for both. You don't mention shutter speed - maybe wind was a factor? Possibly GEM has softened the scan for the 67. Whatever happened, the 67 scan is not as sharp as I would expect. Here's a Coolscan 9000 (standard glassless holder) scan of 120 Fuji Reala 100 film that I've grabbed to illustrate (I probably have sharper ones, but I didn't look further). It's done with ICE, but no GEM, and no sharpening has been done:

 

I'm quite confident in the focus for both images. I evaluated them both from center to corner.
The fence in the foreground is in sharp focus, and the horizon is slightly out of focus. The
crop area seems to be right around the area of sharpest focus. The shutter speed is 1/15th.

I think the apparent softness is just the nature of fine detail in medium to low contrast light.
(notice that the finer grasses to the side of the large clump seems to be in flat light). Your
image is very different. It would be easier to evaluate an image that has approximately the
same scale and lighting contrast as the grasses.

 

Your shot seems to be taken with the sun low, but still the grass seems to be in direct sunlight, which I would have thought would produce reasonably high contrast.

Some other possibilities... Could the focus be off for your scan? The version without GEM you show now doesn't have the hints of grain that one can see in my scan, which could indicate that the scanner wasn't in perfect focus. Or could shutter vibration be a problem? I've heard that that's a concern with the Pentax 67, maybe especially around 1/15s
shutter speed. (My example was with a Mamiya Press, with leaf shutter, and was also done with flash, so there would have been no such problem.)

 

OK. Heres the deal; I download your 5D file, process it, develop it in 16bits AdobeRGB no raw sharpening. Then i apply a 150% Smart Sharpen with a 1 radius using Lens Blur removal. Then i interpolated it using BICUBIC SMOOTHER to a similar size of your crop. I crop it and apply a small amount of print sharpening. Then i put them side by side and reduce both so i can upload it here at a 511pixel wide. I think the result is clear; if you dont know how to process a digital file you end up with a soft image, and trying to do a test without all the knowledge will give you a rong result..like you got. I prefer a test like the luminous landscape that for me seem to have a better understanding of EVERYTHING than from someone who dont, or at least dont seem to. no offence.

 

Thank you for your effort. I must say, however, that this is not a comparison of different
sharpening methodologies for the same raw file, but a comparison of film vs digital?
specifically the 5D vs Pentax 67. I would be useful if you would do a full sharpening routine
on the Pentax file, and compare it to your version of the 5D file. I will email you a small
section of the file if you would be willing to do it.

 

Marty, indeed it is a digital vs digital comparaison..because to be able to compare a film vs digital you need to get the best of both medium before trying to do a comparaison.

You have to be agree that my version of your 5D file is way better, therefore it can be use now as starting point. To be honest i dont have the time to do more conversion this week, but if i had i will be interested of getting both *original* file the one from your scan and the 5D again so i can start from the beginning and than having a real VS result.

But again, i dont have the time this week for it, and the test from the luminouslandscape satisfy me.

thanks for the offer.

 

Oh i must add, from my example you can now see how important it is to have the knowledge to process a digitl file, because if you dont you end up with a image that could be way more than what it is. As Marty says, it was not originaly a post on different methodology of HOW TO SHARPEN, but using wathever method you like, it is important that people understand that a digital image must be sharpen before being printed.

I personnaly discover that using a 2-3 sharpening method was the best, and i was happy to have it comfirmed by Bruce Fraser in is book Real World Sharpenning where he explain the why and how to do it.

I repeat, knowing all you can in both medium is important so you can compare apple vs apple. If you lack on knowledge on one of them you will end up with bad result in your test, and therefore anyone will be able to crush it and prove that your rong.

 

That is fine. It will be interesting to see your result. I must say that your comparing your
sharpened image to my unsharpened image is not relevant. Why did you not compare it to
my sharpened image?

 

because when i donwlod it i didtn see any sharpen version.

 

i just saw it and even your version of the 5D sharpened look unsharped...like it say, i prefer to start fresh form the original because i like to think that being a professional i know what im doing..and it look like it too when i compare mine vs your 5D sharpend. no offence.

 

Why would you have done print sharpening? That seems irrelevant unless you are comparing
print to print.

 

Agree, i rongly use the term. I should have say that a 3 pass sharpening consist in those;

1_ Capture Sharpening is applied early in the image-editing process, and just aims to restore any sharpness that was lost in the capture process.

2_ Creative Sharpening is usually applied locally to accentuate specific features in an image-for example, we often give eyes a little extra sharpness in head shots.

3_ Output Sharpening is applied to files that have already had capture and creative sharpening applied, after they've been sized to final output resolution, and is tailored to a specific type of output process.

So i should have use the term output sharpening to the file, because i resize it.

sorry if that cause some understanding problem : )

 

In my sharpened version, I did a 2 pass process. The first was in DXO Optics, hardly an
unprofessional program. The second pass was with Bruce Fraser's own Photokit Sharpener,
using creative sharpener-high pass 2. I used the same second step with the sharpened
version of the Pentax file. I didn't choose to do output sharpening for the web, and that
certainly contributed to the apparent softness of all images. But, nonetheless, the idea was to
compare apples to apples. Had I done output sharpening it would be comparing oranges to
oranges. I should apply output sharpening to mine so orange is compared to orange.

 

I dont think the uotput sharpening will really change anything here, but feel free to experiement : )

As i say, all step must be fully control so you could compare whatever fruit vs whatever fruit. Lets put your energie to another thing, accepting that maybe you would have to redo some other test that could show something else.

For me the luminouslandscape seem more reliable, not because he say that i prefer digital now, but simply because is test seem more controled.

 

The main advantage of the Luminous Landscape scene over the one I shot is that there is no question of wind being a possible factor, though I believe there was little or no wind in my scene. The other factors are all post-shoot processing.
The comparison is between an image scanned on a Nikon scanner and a digital shot, that's all. This is NOT a test of a drum scan vs digital. I have fully switched from the Pentax 67 to the Canon 5D, and the Pentax is now sold, so I can't do any more tests
Here is a version of the Pentax image for which I have 3 pass sharpening, all with Photokit sharpener. This is what should be compared to your version above. This is oranges compared to oranges. Feel free to submit a third orange.
http://www.americansouthwestphotos.com/comparison/Pentax 67 Gem,
output sharpening.jpg

 

then what was your point if you didtn have the pentax anymore, if you didtn process them correctly to start with, and if you shoot some hay on a maybe or not windy day as to create a test?

what will you want us to understand by it? im not sure anymore what was the purpose of it. Was it to ask for proper digital workflow? to show that you prefer the film image but still you sale it for a reason...

please explain.

 

After all that has been said in this thread about sharpness between the Pentax and Canon
captures, Marty still shows the main difference I keep seeing between film and digital and
that being their color mapping characteristics.

Digital tends to render subjects made up of one color lit with off neutral color temps like
the dead grass in the posted shots with a monochrome feel between highlite and shadow.
This IS an accurate depiction as our eyes perceive the spectral reflectance of similar
objects and textures.

However film whether intended or not by the manufacturer and/or processor is more
about image enhancement. I see four to five color variances making up the Pentax shot to
render the one color of the dead grass-pinky orange in the midshadows of the main clump
of grass, yellowish orange in the highlites, neutral gray in the shadows of the surrounding
grass, maroonish brown in the deepest shadows of the main clump of grass where the
light doesn't hit it.

You have to ask yourself what would you have to do and how much work would be
involved to get the digital to render the same.

 

im sure that with a bit of color balance and hue saturation enhancement all could be achieve pretty easily and quickly.

I dont think film vs digital is relevant anymore, as both of the medium are different, and people could achieve anything with both according you have the rigth knowledge to produce the best file from it. I think the limitation of digital or film is because of the knowledge limitation of the photographer in many aspect. certainly if you go into the scientific debate one could argue that film is better because....but since what really matters is what you could see and like printed and hang on the wall, i dont think its very important to know from what it as been shoot and processed. IMO.

 

My purpose? I was curious to compare, and thought others would be interested also,
which is apparently true based on the number of visits to the comparison page. The
posting is for other people's benefit, not mine. And since I have included sharpened and
unsharpened versions, (the sharpened ones using professional sharpening plug-ins), I
think the results are sufficiently useful because people can choose among the various
versions to compare. Your version added a mango to the mix because you did output
sharpening. Fine, it just means the others have to have output sharpening for a proper
comparison to be made. Your creative sharpening process for the 5D image may be
superior to Photokit Sharpener, which I use, but I didn't want to compare sharpening
methods. That's why I originally posted unsharpened images. Sure digital images need to
be sharpened, but so do scans. I think there is usefulness in having unsharpened images
for comparison.

 

patrick,

That kind of color rendering would require manipulation of the image's cLUT or a trip
through LAB. Not an easy task. I've tried it.

I ended up resorting to gradient mapping and Soft Light blending layers which got me
close but still a bit rough around the edges.

I still prefer digital's immediacey over the cumbersomeness of scanning film or having
others do it with usually unexpected results.

I'm sure some camera manufacturer has a special algorithm included within there camera
or RAW converter that renders color this way. I don't have the ability to find this out
though.

 

"I see four to five color variances making up the Pentax shot to render the one color of the dead grass-pinky orange in the midshadows of the main clump of grass, yellowish orange in the highlites, neutral gray in the shadows of the surrounding grass, maroonish brown in the deepest shadows of the main clump of grass where the light doesn't hit it."

Interesting, Tim, and probably more relevant than the sharpness/resolution questions. I wonder if the 5D shot is handicapped by either its higher contrast relative to color negative film (subtle shadow casts get pushed towards black) or just that this shot is balanced warmer on the 5D. Without full comparison shots it's hard to tell and I don't have time to download the files.

I think it would be fairly easy to at least cool the 5D shadows using split tones in Lightroom.

 

While I completely agree with Patrick's points relative to properly processing images from digital cameras in order to pull out all the quality that's there to pull, I have to point out that the same is true to an even greater extent when it comes to scanning and properly post-processing film images for maximum benefit. Most examples I've seen in these (seemingly endless) film/digital comparisons tend to leave quite a bit on the table on the film side as well.
There have been several very high quality, in-depth, technical evaluations of the differences between film and digital done over the last few years, and the one clear conclusion that can be drawn from the best of these is that it's a very complex issue with few hard and fast "once you get to this point 'x' is clearly better than 'y'" answers.
With all due respect to Marty - given all of the variables involved, all any of these ad hoc comparisons end up showing are the differences for the individual doing the test. They are of very limited usefulness to a wider audience. To attempt to draw any broad meaningful conclusions from a single pair of photographs taken and processed by a single individual under one set of conditions is... well, sorry, but it's just pointless.
Scott

 

Roger,

My suspicions as to why the 5D renders color this way as with all of the digital camera's
samples I've seen is that they don't record light/photons with the same filter effect of film
which records light using layers of color dyes, silver salts all encased in celluloid.

This I believe renders color much like applying polyurethane to dry dull wood.

Also from working with my own Pentax K100D RAW files my other suspicion is based on
the fact that after turning off all color enhancements, the raw preview starts out almost
monochrome and after turning back on the color enhancements chroma noise can be seen
all over zoomed in on detail as done here. There is a filter in RAW Developer, my current
app, called Chromalogic that when applied gets rid of this noise but renders color the
same way as seen in the D5 shot. The same effect can be seen applying heavy Gaussian
blur to a Color Blend mode layer in PS.

 

Saddly, i think Tim that you didtn have the chance to see a good digital file in your life, that sad. I hope one day you come across one, and you realize how good it could be, when properly exposed, developed and processed. Its amazing how in 2007 people seem to fight the idea that digital could be equal or superior than film IF well produce.

Top photographer around the world must be totally mad to shoot with them then according as you say that digital image are dull and lifeless.

i wonder why? Anyway, im not interested in a longer debate in that area, as i already know how the outcome will be.

Bottom line is, if one prefer to go with film, and another prefer to go with digital..let it be, and live happy with your choice : )

 

The Pentax 67 image does not look shrp, for whatever reason. There seems to be a problem.

 

Here's a before and after RAW Developer screenshot of one of my K100D PEF's illustrating this monochrome feel. I mistook the name of the filter to be ChromaLogic. It's called Color Smoothing. ChromaLogic does something else I can't figure out yet. And Roger your suggestion of a different contrast curve applied to negative film is probably true because I had to severely apply more contrast to the original RAW file to bring out all the multicolors in the grass which amplify's the chroma noise in the limestone. The RAW file without Pentax's secret sauce incamera processing tends to render pretty flat with off looking color temps in RAW Developer.

 

patrick,

I've seen a sample close-up shot of a section of a Van Gogh painting that was mind
blowing captured by a $12,000 scanback from a company out of Spain who restores
museum masterpieces. It recorded the finely ground pigment specks embedded in the
paint without any noise or loss of detail.

We're talking here about a $3000 camera and how it records textured detail off in the
distance.

 

Robert, regarding the Pentax image not looking sharp, remember it is a small section of the
image. See the comparison page to see the uncropped image. The link is in the first post of
this thread.

 

Marty,

Still, even though the Pentax image is a small section of the image, I'm worried about its
sharpness. Shouldn't you be able to resolve the grain of the film? I can definitely resolve
the grain on my negative scans on my Nikon LS-4000...

Unless your film is really that grain-less

Point being... it seems to me you could get a better quality scan... in support of film, hehe.
But I don't want to get into that debate either. Both have their advantages.

Rishi

 

Tim,

But, regardless of that monochromatic feel on the small scale after color smoothing, I
don't think the perception of color in the overall image, viewed zoomed out, is hurt.

And, as much as I love film, the 'cleanliness' of digital images seems to lend to better
sharpening as well as an image that 'appears more pristine & sharp' to the eye...

In my experience...

Also, interesting discussion about the way colors are recorded in film vs. digital.
Remember the Bayer pattern in digital... these filters on the CCD pixels have rather narrow
(and sometimes strange) transmission spectra... which I'm sure leads to some nuances in
digital capture. Of course, a lot of this can be corrected by a good RAW converter
(typically, the manufacturer's!).

But this has got me thinking about the structure of film... I forget -- are there filters in
between the red, blue, and green sensitive layers in film? If not, if the absorbance spectra
of the dye couplers are largely what determine the color responses, I'd imagine one could
more accurately capture colors the way our eyes perceive in film (with the correct
chemistry... i.e. dye couplers with absorbances matching those of our cones) vs the Bayer
pattern CCD.

Actually, this warrants a good dive into my color/film theory books, which have been
gathering dust on my shelf

Cheers,
Rishi

 

I expect to see grain in the crop. It's not there, which seems odd. How does the negative
look on a light table with a loupe? If it's sharp, it could be a scanner focus problem.

 

Exactly, Robert. For example, note the obvious grain in the Fuji Reala scan above...

 

OK, I looked up the structure of film and there's only a yellow filter in the film, underneath
the top-most blue-sensitive layer. The subsequent two layers underneath the yellow filter
are blue sensitive, but the yellow filter filters out the blue light, only allowing green & red
light to pass to the bottom two layers. The first these bottom two layers is sensitive to
green and the bottom-most layer is sensitive to red (via dye couplers). No other inherent
sensitivity in these two layers whatsoever.

So, back to the reason I brought this up: seems to me that, due to the rather wide
absorption spectra of the dye couplers vs. the narrow transmission bands of the Bayer
pattern filters (at least the spectral data I've seen for some of the Bayer pattern CCDs on
digital cameras), film might have the ability to get closer to emulating the response of the
photoreceptors in our own eyes... but that's just a guess, as I haven't studied the
absorption spectra of dye couplers in film... yet

But, again, such color issues can presumably largely be resolved by appropriate RAW
converters. Otherwise we wouldn't get the stunning images digital capture affords us.

Darn, I just realized I dragged myself into this film vs. digital debate, and without any
instigation at that!

Fun, anyhow, but perhaps better suited for a different thread.

So I take my leave... Rishi

 

Regarding the apparent absence of film grain, digital GEM was used for the above Pentax 67 scan. So that obviously begs the question did GEM soften the image? Well, there is a slight softening. Here is the image with no GEM.
http://www.americansouthwestphotos.com/comparison/Pentax 67 No GEM
sharpened.jpg

 

Marty -- much better!

 

I looked at Tim's color smoothing example page:
http://www.photo.net/bboard/big-image?bboard_upload_id=40987784

I use Lightroom for RAW conversions (same as ACR)
If you turned off color noise reduction wouldn't you get a version like the one on the left where different color "noise" helps form the image? Are you happy with the image on the left vs film?

I usually dial all noise reduction down very low myself to preserve image detail.

 

Re: Lightroom for RAW conversion of Canon RAW files... I've found LR to do a horrible job
with Canon Rebel XTi RAW files... rendering rather 'monochromatic' feeling images, with
large hue shifts for oranges/reds especially. I believe this has been acknowledged before.

Digital Photo Professional (Canon's own software) does a much better job of the RAW
conversion, although with the ACR Calibrator script and a ColorChecker card, one can get
pretty decent colors in ACR/Lightroom. But not as good as in DPP, in my tests.

I think the ultimate workflow would involve DPP for conversions, importing into LR, then
applying camera calibration (R, G, B primaries) settings as determined by manually
performing a calibration on the DPP converted image of a ColorChecker card (manual
process described in Real World ACR, the latest version, by Jeff Schewe and Bruce Fraser)...
again, that's for another thread.

Rishi

 

The image on the left looks like it has chroma noise due to improper white point estimation.

 

All i could add is;

Everytime i heard someone complain about raw development problem, color noise problem, digital rendering etc.. most of the time (if not everytime) it was a matter of lack of knowledge in this field.

As for example;

someone using a plugin to reduce the noise on a file and complain for the lack of sharpenes..when eveybody know that if your using something to remove the noise, the image will indeed get softer.

or someone using a colornoise reduction plugin and complain that is file look like a monochrome after with color lost and no vibrancy...

or not sharpening a digital file or a image from a film and saying that the image is soft...

basically what i read here : )

That why i dont think this test is relevant, and thats why i think that before one could do a test like that it should be able to control every aspect of BOTH technology. As another readers says,

...< any of these ad hoc comparisons end up showing are the differences for the individual doing the test. They are of very limited usefulness to a wider audience. To attempt to draw any broad meaningful conclusions from a single pair of photographs taken and processed by a single individual under one set of conditions is... well, sorry, but it's just pointless...>

 

Patrick,

I wouldn't say it's lack of knowledge in the field *everytime*. I agree digital has immense
potential, it's great, but quality of RAW conversion depends on how well the RAW converter
is able to translate RAW data to real colors. Since some developers build these converters
themselves (e.g. Adobe Camera Raw) by shooting targets with camera models, they may or
may not agree with manufacturer RAW converters, and that may or may not be a problem
depending upon which RAW converter provides more accurate conversions.

Using Gretagmacbeth ColorCheckers, for example, can give you more accurate results
after calibrating, but they, in my experience, can't always compensate for a good RAW
conversion done by an accurate RAW converter (for that particular camera).

In particular, I've found LR/ACR's conversions of Canon RAW files to be unreasonable in
certain colors, as have many photogs. Adjusting the camera calibration brings the result
near the color conversion provided by Canon's DPP, but with other artifacts/problems
abound. For example, in my experience, magentas get over-saturated upon this camera
calibration correction.

Cheers,
Rishi

 

Oh, so this is where the thread was moved to...Casual Conversations. Thought it got
dropped through some database glitch.

Anyway, I posted the RAW Developer Color Smoothing image only to illustrate what MIGHT
be the cause for why digital camera's render color this way. Just throwing ideas out here
and see what sticks. I'm not pushing a digital vs film argument either. I just see a
difference.

Anyone else who can show a DSLR can render color in distant detail the same as film post
it here and indicate how you did it.

 

"In particular, I've found LR/ACR's conversions of Canon RAW files to be unreasonable in certain colors, as have many photogs. Adjusting the camera calibration brings the result near the color conversion provided by Canon's DPP,"

How do you find these compare to in camera Jpeg conversions? With tweaked presets I get nearly identical results between my 20D RAW files in Lightroom and in-camera jpegs. It's a decent starting point for further edits I think.

 

In camera JPEG conversions are very similar to Canon DPP's processing of RAW... I assume
they're using a similar engine in-camera for JPEG processing.

As you say, 'with tweaked presets' you get nearly identical results in LR as in-camera
JPEGs. My 'tweaked preset' in LR, as spit out by ACR calibrator script run on a RAW image
of Gretagmacbeth ColorChecker:

Camera Calibration:

Shadows: +2
Red Primary:
Hue: -31
Saturation: +30

Green Primary:
Hue: -26
Saturation: +30

Blue Primary:
Hue: +8
Saturation: +3

That seems like quite a bit of tweaking to me! Colors look good, if not great, after these
calibration; however, sometimes I run into problems. For example, the magenta/pink label
on a container of iodized table salt ended up being gaudily saturated magenta in LR after
application of these presets, whereas in the DPP conversion, it was much more reasonable.
All other colors remained largely the same between the DPP conversion and the LR
conversion (after the LR calibration).

I like DPP's presets for daylight balancing, etc... for example 'Natural' and 'Landscape'.I
find their color conversion better, naturally, probably b/c they're the manufacturer of the
camera, they understand it well, they've worked with it for longer than Tom Hogarty over
at Adobe (I believe he's the one that builds the camera profiles) has had the chance to...
right?

Overall, though, I find Lightroom to be an ingenious piece of software. Hence why I export
tiffs from DPP and then import them into LR for all subsequent post-processing.

I could post some results of the gaudy magenta/pink example... in a bit.

Cheers,
Rishi

 

Hmm. I haven't seen these issues, but I've only tried LR and didn't even install DPP. As with my film scanning I shot a grey/white/black target under daylight and used that to create reference presets for nominal exposure, over, and under.

Is magenta the only thing to watch out for? I would appreciate an example.