Ad
Ad
Ad
Pages: [1] 2 3 ... 15 »   Bottom of Page
Print
Author Topic: Can you see a difference in small prints between an MFDB and a DSLR?  (Read 17600 times)
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7635


WWW
« on: February 27, 2014, 12:03:14 PM »
ReplyReply

Hi,

The question has been posted on a separate thread. My experience is essentially that I see no difference between my P45+ and my Sony SLT99 in A2 size prints, but I would think I see differences in larger prints.

I got some fierce suggestions that I use inferior technique, so I took a somewhat different approach. This time I have used a test shot from the Pentax 645D at Imaging Review and compared with my SLT99. Both images were developed in LR5.3 with identical settings and analysed using Imatest.

Imatest has an option for calculating SQF (Subjective Quality Factor) that is based on research by Ed Granger at Kodak Eastman. If we assume 17" print height at 40 cm viewing distance the SQF figure for the Pentax 645D would be around 93 on the Pentax 645D and around 90 on the Sony Alpha 99. According to Kodak research it takes about 5% of SQF to make a visible difference.

So, according to Kodak research there would be no discernible difference on an A2 print (17" wide) between a Pentax 645D image shot at Imaging Resource and my SLT 99 shot when viewed at 40 cm distance.

Looking at images at 100 cm height there would be a perceptible difference, still viewed at 40 cm distance.

Imatest also calculates an optimal sharpening. This probably corresponds to something like an extra sharpening step with radius = 2 and an amount of 15%.

Best regards
Erik



« Last Edit: February 27, 2014, 12:28:37 PM by ErikKaffehr » Logged

KLaban
Sr. Member
****
Offline Offline

Posts: 1677



WWW
« Reply #1 on: February 27, 2014, 12:15:21 PM »
ReplyReply

Erik, if your experience is essentially that you see no difference between your P45+ and your Sony SLT99 in A2 size prints then use your Sony for the A2 prints and your P45+ for your larger prints.

Above all enjoy making images rather than using your cameras as testbeds.

I genuinely wish you well and good shooting!
« Last Edit: February 27, 2014, 12:24:41 PM by KLaban » Logged

Fine_Art
Sr. Member
****
Offline Offline

Posts: 1094


« Reply #2 on: February 27, 2014, 12:31:11 PM »
ReplyReply

http://www.imatest.com/docs/sqf/

What is the sqf rating of the IQ250/260 shots in imatest?

What is the correlation of MP to print size? Your 24MP gives you a 90 at 17" diagonal, what about 16MP (90), 36MP (90)?
Logged
EricWHiss
Sr. Member
****
Offline Offline

Posts: 2427



WWW
« Reply #3 on: February 27, 2014, 12:38:38 PM »
ReplyReply

Wait, you mean you are going by numbers and haven't made prints? Is this a joke?  Where are you getting the SQA numbers from anyhow?  Are you familiar with all the variables in Imatest that affect SQF besides the print size? The SQF will be different for each lens and aperture tested and there can be much larger differences than 5% between two different lenses on the same system.  Therefore your approach will not be convincing.
 
What's your goal now?  Some kind of extra "measurbation"?

There are many qualities besides sharpness that separate MFDB images from DLSR images. Forget sharpness for a moment (which is where I think you are going with SQF) The format size differences alone make a visible impact and also the way the images rolls from in focus to OOF areas is different.    

If you can't see these kinds of "forrest through the trees" things then well… I don't know but maybe you waste your time and ours.
Logged

Authorized Rolleiflex Dealer:
Find product information, download user manuals, or purchase online - Rolleiflex USA
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7635


WWW
« Reply #4 on: February 27, 2014, 12:39:48 PM »
ReplyReply

Hi Keith,

Someone was asking about the visible difference between 17" prints on another thread. This is a response to that posting

What this posting shows is what to expect based on research at Kodak Rochester comparing a 39 MP back with a 24MP DSLR using identical processing. The 39 MP image is none of mine in this case, it is from one of the tests at Imaging Resource.

I would really recommend anyone considering an MFDB to download a reference image taken with their cameras of choice and do large size prints.

Regarding your suggestion to use the SLT 99 up to A2 and the P45+ for larger prints, it would suggest that print size would be known at exposure, and that is often not the case.

Best regards
Erik



Erik, if your experience is essentially that you see no difference between your P45+ and your Sony SLT99 in A2 size prints then use your Sony for the A2 prints and your P45+ for your larger prints.

Above all enjoy making images rather than using your cameras as testbeds.

I genuinely wish you well and good shooting!

Logged

Paul2660
Sr. Member
****
Offline Offline

Posts: 1900


WWW
« Reply #5 on: February 27, 2014, 12:45:05 PM »
ReplyReply

No in small prints I can't.  Where I will start to see the difference is in a print that is 23 x 33 (fits a 30 x 40 frame matted) and larger.  I like to print up to 36 x 72 and I for this I prefer not to interpolate unless totally necessary.  Here the larger resolution of the MFD helps as you can get there much easier.  However I realize you can stitch 35mm, as I have been doing that since early 2003.

As for small prints no, I can't see much difference, say 20 x 30 and smaller.  However it's also sometimes depending on the subject matter of the print.  If the image has a lot of small details the MF back print will do better in my experience. 

This is printing on a 9900 only.  Canvas or paper.

Paul C.
Logged

Paul Caldwell
Little Rock, Arkansas U.S.
Photography > http://photosofarkansas.com
Blog> http://paulcaldwellphotography.com
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7635


WWW
« Reply #6 on: February 27, 2014, 12:56:03 PM »
ReplyReply

Hi Eric,

You may perhaps realise that other people than yourself are entitled to an opinion that may differ from yours?

I did make prints, but they are always affected by subjective evaluation. SQF numbers are much less sensitive to lens variations than you may believe. I have every reason to believe that the image I used from Imaging Review is correctly taken. They use focus bracketing and they do testing for living.

I did measure SQF with Imatest, a program you also use and I am quite familiar with, since I was using it since it's inception.

Roger Ciala does at lens rentals uses Imatest extensively, as they test all incoming lenses from rentals. He is quite clear on that significant variation in Imatest results like LW/PH result in little visible difference in prints.

It is difficult to post prints over the net, as you may know. I would recommend anyone investing a major amount into a new system to download raw images from a source, like Imaging Resource, do they own processing and printing. I have done that for all systems I bought and some others ones, too.

Best regards
Erik

Wait, you mean you are going by numbers and haven't made prints? Is this a joke?  Where are you getting the SQA numbers from anyhow?  Are you familiar with all the variables in Imatest that affect SQF besides the print size? The SQF will be different for each lens and aperture tested and there can be much larger differences than 5% between two different lenses on the same system.  Therefore your approach will not be convincing.
 
What's your goal now?  Some kind of extra "measurbation"?

There are many qualities besides sharpness that separate MFDB images from DLSR images. Forget sharpness for a moment (which is where I think you are going with SQF) The format size differences alone make a visible impact and also the way the images rolls from in focus to OOF areas is different.    

If you can't see these kinds of "forrest through the trees" things then well… I don't know but maybe you waste your time and ours.

Logged

KLaban
Sr. Member
****
Offline Offline

Posts: 1677



WWW
« Reply #7 on: February 27, 2014, 01:05:44 PM »
ReplyReply

Regarding your suggestion to use the SLT 99 up to A2 and the P45+ for larger prints, it would suggest that print size would be known at exposure, and that is often not the case.

Erik, that's a persuasive argument for using your P45+ wherever possible ;-)

Best

Keith
Logged

tho_mas
Sr. Member
****
Offline Offline

Posts: 1696


« Reply #8 on: February 27, 2014, 01:10:10 PM »
ReplyReply

Someone was asking about the visible difference
about the visible difference... not the measurable difference. Do you think your post is really helpful to answer the initial question?
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3748


« Reply #9 on: February 27, 2014, 01:29:36 PM »
ReplyReply

about the visible difference... not the measurable difference. Do you think your post is really helpful to answer the initial question?

Hi Thomas and others,

The SQF metric that Erik used, is the quantification of the visual/subjective image quality differences. When the SQF tells that there is not a significant difference, then there is no relevant visual/subjective difference.

An objective metric tends to allow a more meaningful discussion. If there is no significant quantitative difference, and people state that they do see a difference, there is a possibility for more targeted discussion, e.g. about viewing distances or print pipeline optimizations that can explain such differences. It took me e.g. quite some effort to convince some of the establishment here at LuLa, that printing at 600/720 PPI does make sense for higher quality output (Qimage users already knew that for more than a decade). First met with skepticism, now the official recommendation.

Cheers,
Bart
Logged
jerome_m
Sr. Member
****
Offline Offline

Posts: 621


« Reply #10 on: February 27, 2014, 01:35:58 PM »
ReplyReply

Imatest has an option for calculating SQF (Subjective Quality Factor) that is based on research by Ed Granger at Kodak Eastman. If we assume 17" print height at 40 cm viewing distance the SQF figure for the Pentax 645D would be around 93 on the Pentax 645D and around 90 on the Sony Alpha 99. According to Kodak research it takes about 5% of SQF to make a visible difference.

So, according to Kodak research there would be no discernible difference on an A2 print (17" wide) between a Pentax 645D image shot at Imaging Resource and my SLT 99 shot when viewed at 40 cm distance.

Erik, I said it once and I will repeat it: if one tests two cameras and equalises per pixel resolution, uses an aperture that will equalise lens differences, makes sure that light and exposure is within the usable values of the two cameras, uses subjects with simple, pure colours and only shoots resolution targets, one will find that the two cameras are equivalent for any couple of cameras. But that is not a limitation of the cameras, it is a limitation of the test design.

That test from Kodak is not an answer to your question: "Can you see a difference in small prints between an MFDB and a DSLR?". That test from Kodak is an answer to the question: "At what size can we print pictures taken in ideal conditions with a MFDB and a DSLR and have a standard observer not notice the imperfections of the camera?". It is a completely different question and the reason that you are increasingly frustrated with the discussion is because you are mistaking one question for the other.

You could notice the difference in small prints between your MFDB and your DSLR if you were shooting imperfect pictures, because the imperfections of the cameras are different. Can you try to do that?
Logged
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7635


WWW
« Reply #11 on: February 27, 2014, 01:57:19 PM »
ReplyReply

Hi,

A much better answer than bragging. Look here, you cannot send prints on the web. Folks can download raw images from the net and do their own processing and printing, that is what I would recommend, but there are not that many comparable images on the net.

The figures here are essentially based on long research by and calculated by industrial quality tools. The Pentax shot I compare with was made under optimal conditions by professional testers.

In my experience, the differences are not readily observable, but I had only a few observers. Quite a few MFD users indicate that there is little difference between the new 36 MP cameras and low end MFD to which my P45+ belongs. Opinions differ.

Another reflection is that you start with an image, and do capture sharpening, after that there may be some creative sharpening. The image will than be resampled to printer resolution that may be 360PPI or 720PPI on Epsons and output sharpening applied. Different pipelines may give different results, but any image will go trough many conversions before it is converted to small dots on paper.

Jeff Schewe and the late Bruce Raser have written a small book on sharpening, Jeff writes that research by Bruce indicated that 180 was plenty of resolution for prints viewed by 20/20 vision at 50 cm or more. That resolution would correspond to 22 x 33 inch on a 24MP DSLR clearly larger than the 17" that was asked.

Page 30 in that book shows smallish prints from iPhone 4S, Fuji Fine Pix A820, Canon EOS Digital Revel XT, Canon EOS 1DsIII and a Phase One P65+. Jeff says that the images cannot be told apart. Of course, the P65+ image could be enlarged to 28"x37" and the iPhone 4s to 4"x5.3", but what Jeff says is that when the image is good enough there will be little visible difference.

A while ago, there was a conversation between Mikael Reichmann and the famous printer Ctein. Ctein used Pentax 67 in former days but now shoots 4/3, which he clearly says is good enough. Good enough for A2 size prints which is the largest he makes.

Best regards
Erik

about the visible difference... not the measurable difference. Do you think your post is really helpful to answer the initial question?
« Last Edit: February 27, 2014, 02:15:47 PM by ErikKaffehr » Logged

bpepz
Full Member
***
Offline Offline

Posts: 202


« Reply #12 on: February 27, 2014, 02:00:54 PM »
ReplyReply

Something else to consider is that most of the "MF" look has to do with bokeh and depth of field rendering.  And no, I am not talking about how a 80mm 2.8 is supposed to look like a 50mm 1.4 on FF, it comes close, but there are big differences besides DOF thickness. A fast 1.4 lens on FF will have more blur then even f/2 medium format glass most of the time, BUT, there is something too it that makes the images much more 3D. It is almost like someone used clarity on the entire image, but without the harshness. It is something you do not get on 35mm. Shooting test targets or subjects at infinity, shooting stopped down, are not going to show you much.

Even if you equalize aperture, focal length and such, there is one factor that you cannot equalize, no matter how many focal reducers or combinations you use. It is magnficaiton. The larger the sensor area you are using, the larger the magnficaiton you will have to use to get the same framing, this is regardless of focal length. On 8x10 just a simple portrait with a "normal" lens is putting you close to 1:1. So it would make sense that shooting at infinity or small apertures you would see zero difference.
Logged
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7635


WWW
« Reply #13 on: February 27, 2014, 02:08:52 PM »
ReplyReply

Hi Jerome,

Yes I understand what you say, sort of. It is very difficult to compare different images. Admittedly I like numbers. I am of course aware that numbers are not the best answer.

Best regards
Erik

Erik, I said it once and I will repeat it: if one tests two cameras and equalises per pixel resolution, uses an aperture that will equalise lens differences, makes sure that light and exposure is within the usable values of the two cameras, uses subjects with simple, pure colours and only shoots resolution targets, one will find that the two cameras are equivalent for any couple of cameras. But that is not a limitation of the cameras, it is a limitation of the test design.

That test from Kodak is not an answer to your question: "Can you see a difference in small prints between an MFDB and a DSLR?". That test from Kodak is an answer to the question: "At what size can we print pictures taken in ideal conditions with a MFDB and a DSLR and have a standard observer not notice the imperfections of the camera?". It is a completely different question and the reason that you are increasingly frustrated with the discussion is because you are mistaking one question for the other.

You could notice the difference in small prints between your MFDB and your DSLR if you were shooting imperfect pictures, because the imperfections of the cameras are different. Can you try to do that?
Logged

ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7635


WWW
« Reply #14 on: February 27, 2014, 02:23:46 PM »
ReplyReply

Hi,

My guess is that it is much easier to make an f/2.8 lens than an f/1.4 lens. Most f/1.4 lenses have significant colour fringing for instance. Zeiss made an "all out" effort to make a near perfect 55/1.4 lens designing the Otus. That is a lens essentially free of colour bokeh. An 80/2.8 can perform decently well without stopping down at all.

That kind of differences would be visible regardless of print size.

My guess is that it was not kind of difference the OP was asking for, but yes you have a very good point.

Best regards
Erik

Something else to consider is that most of the "MF" look has to do with bokeh and depth of field rendering.  And no, I am not talking about how a 80mm 2.8 is supposed to look like a 50mm 1.4 on FF, it comes close, but there are big differences besides DOF thickness. A fast 1.4 lens on FF will have more blur then even f/2 medium format glass most of the time, BUT, there is something too it that makes the images much more 3D. It is almost like someone used clarity on the entire image, but without the harshness. It is something you do not get on 35mm. Shooting test targets or subjects at infinity, shooting stopped down, are not going to show you much.

Even if you equalize aperture, focal length and such, there is one factor that you cannot equalize, no matter how many focal reducers or combinations you use. It is magnficaiton. The larger the sensor area you are using, the larger the magnficaiton you will have to use to get the same framing, this is regardless of focal length. On 8x10 just a simple portrait with a "normal" lens is putting you close to 1:1. So it would make sense that shooting at infinity or small apertures you would see zero difference.
Logged

tho_mas
Sr. Member
****
Offline Offline

Posts: 1696


« Reply #15 on: February 27, 2014, 02:29:19 PM »
ReplyReply

Hi Thomas and others,

The SQF metric that Erik used, is the quantification of the visual/subjective image quality differences. When the SQF tells that there is not a significant difference, then there is no relevant visual/subjective difference.
:-) maybe. maybe not. I don't know. I don't care about those comparision tools.
What I do know is that this forum is completely polluted with too much theoretical talk (scientific and pseudo scientific talk). In every thread I get explained that my DBs show aliasing, moire and pseudo details (as f I wouldn't know it...). 2 posts later someone talks about AA filters and deconvultion sharpening. In the third post a completely unexperienced user finds out that C1 doesn't produce accurate colors (based on a color checker passport shot with mixed lights). In the 5th post someone is talking about skin tones although he never shoots people. But he is finding flowers and/or animal hair compares quite good to skin. And in the 6th post we talk about the D800 or D800E or the A7R.
In EVERY thread, 3 times a day.
That's all fine... but it gets really, really old. Not a particular issue is getting old... but the way and the style things get discussed here. Erik invested 15K just to be able to talk about MFD. Now, this is an issue that should be treated seriously... I really think he needs professional help. But, please, let's not discuss this on this forum...
Bart, sorry, maybe the method implemented in the said SQF thing is great. But I simply don't care (anymore).
Logged
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7635


WWW
« Reply #16 on: February 27, 2014, 02:59:30 PM »
ReplyReply

Hi,

I buy that argument! But I don't think that was the original question.

Best regards
Erik

Something else to consider is that most of the "MF" look has to do with bokeh and depth of field rendering.  And no, I am not talking about how a 80mm 2.8 is supposed to look like a 50mm 1.4 on FF, it comes close, but there are big differences besides DOF thickness. A fast 1.4 lens on FF will have more blur then even f/2 medium format glass most of the time, BUT, there is something too it that makes the images much more 3D. It is almost like someone used clarity on the entire image, but without the harshness. It is something you do not get on 35mm. Shooting test targets or subjects at infinity, shooting stopped down, are not going to show you much.

Even if you equalize aperture, focal length and such, there is one factor that you cannot equalize, no matter how many focal reducers or combinations you use. It is magnficaiton. The larger the sensor area you are using, the larger the magnficaiton you will have to use to get the same framing, this is regardless of focal length. On 8x10 just a simple portrait with a "normal" lens is putting you close to 1:1. So it would make sense that shooting at infinity or small apertures you would see zero difference.
Logged

jerome_m
Sr. Member
****
Offline Offline

Posts: 621


« Reply #17 on: February 27, 2014, 03:05:13 PM »
ReplyReply

Hi Jerome,

Yes I understand what you say, sort of. It is very difficult to compare different images. Admittedly I like numbers. I am of course aware that numbers are not the best answer.

I don't think that you quite understand what I say. There is nothing wrong with numbers. I like numbers as well. I am actually a trained engineer and mathematician.

My guess is that it is much easier to make an f/2.8 lens than an f/1.4 lens. Most f/1.4 lenses have significant colour fringing for instance. Zeiss made an "all out" effort to make a near perfect 55/1.4 lens designing the Otus. That is a lens essentially free of colour bokeh. An 80/2.8 can perform decently well without stopping down at all.

That kind of differences would be visible regardless of print size.

Yes, that is one thing. An f/2.8 lens will have considerably smaller optical aberrations, most noticeably chromatic and spherical aberrations. This directly influences depth of focus. The formulas that everyone uses to estimate depth of focus suppose that the aberrations are negligible and they are not for a f/1.4 lens.

Erik invested 15K just to be able to talk about MFD.

I think that Erik spent 15K to be able to understand the difference between MFD and smaller formats. That is a completely different perspective and, in my opinion, entirely legitimate.

I think that because this is what I did. I wanted to understand what the difference was and where it came from. Having used the camera, I found where some (maybe most) of the differences came from and, as a consequence, I find that MFD are wonderful cameras. I am glad I spent the money. Can money buy anything more precious than experience and knowledge?
Logged
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7635


WWW
« Reply #18 on: February 27, 2014, 03:24:32 PM »
ReplyReply

Hi,

If you are not interested in a thread you can just ignore it.

Regarding the 15K investment I could do with significantly smaller investment if I just wanted to talk MFD, but I also invested in five lenses and a flexbody. That stuff is for taking pictures of which I have posted quite a few. But yes, I feel sometimes I would need a shrink, but not a great idea to take a lot of money with you in the grave, it won't serve you on the other side.

Most of my recent shooting is with the blad and I enjoy it, at least for the time being.

Best regards
Erik

:-) maybe. maybe not. I don't know. I don't care about those comparision tools.
 Erik invested 15K just to be able to talk about MFD. Now, this is an issue that should be treated seriously... I really think he needs professional help. But, please, let's not discuss this on this forum...

Logged

EricWHiss
Sr. Member
****
Offline Offline

Posts: 2427



WWW
« Reply #19 on: February 27, 2014, 03:25:43 PM »
ReplyReply

Erik,
I needed to work more with my Mac Beta version of Imatest and I have taken test images with most all of the lenses in Rollei 6000/Hy6 mount so I ran some of them through to look at SQF numbers.

First of all, as I mentioned there are quite a lot of variables that influence SQF results - so YMMV.   There's the lens of course, then the aperture, then the region of interest chosen (center or edge), then capture level sharpening if used, noise reduction,

My recollection was most of the Rollei lenses scored in the high end, but I didn't realize that they were as high as they were.   I'm looking at the AF 80mm Xenotar lens for Rollei 6000/Hy6 @ f/8 data now.  Every point in the frame is scoring over 98 at a picture height of 40cm and most of the center is over 99.   I'm attaching one chart from a test point on the very left edge just for reference which reads almost 99 at picture height of 40cm.  The center points are all ranging from 99.8 to 99.3
  
Erik - Hopefully from this actual test datum - you'll begin to understand why I have always felt you could be getting more from your MFDB.  It's true that the Rollei lenses are quite excellent, but I believe that even your older Hasseblad should be capable of more than you have shown.   

Please refrain from asking me to supply you RAW files for your own enjoyment.  At some time this year I will be publishing comprehensive lens test data on all the Rollei lenses for Hy6 and 6000 series cameras.

I have no idea where you got your SQF numbers for your Pentax 645D and other cameras but you should really check to make sure what parameters were used so that the can be compared fairly in this hypothetical test of yours.  And as pointed out - if the question is whether a person could see a difference in a print, then actually using real prints is probably the best approach.


 


« Last Edit: February 27, 2014, 03:31:14 PM by EricWHiss » Logged

Authorized Rolleiflex Dealer:
Find product information, download user manuals, or purchase online - Rolleiflex USA
Pages: [1] 2 3 ... 15 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad