Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: feppe on September 09, 2007, 10:46:02 am
-
Based on the lively discussion (http://luminous-landscape.com/forum/index.php?showtopic=19198) on another thread, I came up with the idea to have a quiz on whether LL readers can really tell the difference between a MFDB and dSLR shot based on a web-sized JPEG.
So I spent the Sunday afternoon on PhotoSig, PBase, photo.net and flickr, collecting images from both formats. The results is a quiz with a wide variety of 50 MFDB and 50 dSLR shots, with various subject matters, level of expertise, etc. The quiz takes 10-20 minutes, so it's not as bad as it sounds.
So, let's see people put their money where their mouth is and get answers to the questions. Can you tell the difference on unfamiliar photographs? Can you pass the quiz (75% correct)? Will you dare to post your results here? Who will be the King of the Backspotters? And perhaps most importantly: is there any point in this?
Well, here's the quiz (http://s-9i160-15679.sgizmo.com/). Enjoy!
-
Well, I got 64%. There is just no way in hell you can judge from pics that size. Besides, a dslr in the hands of a good photographer will produce better results than a DB in the hands of a crappy one any day of the week, and most of the pics in that survey were a sorry sight. I've made amazing 70x100cm enlargements from my 1Ds, but I can also honestly say that a perfect shot from my P25 will make any dslr green with envy. The detail and depth is almost scary at times. This will of course only make a good photo better, nothing more. In the end it's always the image that is important, and no one will be a better photographer just by buying a DB.
-
The pictures are too small and to compressed.
-
Strange: I also got 64%.
-
Strange: I also got 64%.
[a href=\"index.php?act=findpost&pid=138201\"][{POST_SNAPBACK}][/a]
62%
-
Hahaha,
I got no more than 47% but indeed it is very difficult to assess anything based on those small JPG's. To me most could have been DSLR and some even cell phones.
-
60% correct.
I have to judge on 3D look with small pictures and a lot of pictures are just not that good lit to start with.
On the landscapes you will see some pictures looking VERY flat and some looking stunning 3Dimensional.
-
59%,,,but I judged a lot of them on their crop!
-
60% correct.
I have to judge on 3D look with small pictures and a lot of pictures are just not that good lit to start with.
On the landscapes you will see some pictures looking VERY flat and some looking stunning 3Dimensional.
[a href=\"index.php?act=findpost&pid=138212\"][{POST_SNAPBACK}][/a]
Wait a second....Are you now saying its the lighting and not the back that is causing the "3d" look?
If these backs were so far superior to the dslr, that "3d" look should be visable regardlees of the quality of the light.
So which is it, the back or the "light"?
-
Your premise is ridiculous. Judging image quality on web-sized jpegs????
Why not judge them based on verbal descriptions?
-
Here are some interim results after 50 completed surveys:
68% of respondents have more than 5 years photography experience
76% are professionals
Of those shots that the responders were unsure of (40-60%):
- 20 were dSLR shots
- 20 were MFDB shots
Of those photos which were overwhelmingly (75+%) voted either dSLR or MFDB:
- Those shots which were voted to be dSRL: 11 correct, 1 incorrect
- Those shots which were voted to be MFDB: 5 correct, 5 incorrect
I'm not drawing any conclusions at this moment, although the latter pairing is quite interesting. I have quite a bit of bandwidth left so I'll keep this on until I hit 90% or so of my bw for the month - if I do. Hmm, seems unlikely: my bw can take 3000 completed surveys, so spread the link far and wide!
Please note that the images are culled from the web from various sources, so I have no access to originals or full-sized crops, so what you see is what you get
-
The pictures are too small and to compressed.
They're no smaller than some of the images posted to show the "3Dness" of MFDBs in the other thread.
-
Your premise is ridiculous. Judging image quality on web-sized jpegs????
The claim was made by several MFDB shooters that the difference between MFDB and DSLR images was distinguishable even in web-sized JPEGs, and they even posted samples to "prove" this. Read the thread cited in the first post and the point of all this will make more sense.
-
i would like to thank you officially for showing what really is a no-brainer anyway...you cannot judge anything from a jpeg on the web...even if it was much bigger....there is so much more to this...and a very well shot, well manipulated DSLR file can beat a poorly shot DMFB file any day...even in print....
some of the landscape shots look like they are layered exposures which would make it almost impossible to tell from a print (as long as the mpix count is close)...
btw i got 51% and i pretty much based it on what format i thought the guy behind the camera thought he should have at that point...
-
It seems like a few people here are warming up for war, while others like me are tired, specially after the almost 15 pages of the other thread. So I'll write the same as in the other one.
Time to close the door.
-
Hi,
Maybe you don't get how depth works (I don't mean disrespect).
Let's make it very simple.
When you photograph a white piece of paper lying on the floor front straight above no human soul can see the difference between a DSLR, P&S, MF etc.
HOWEVER
When you go in the forest in the goldenhours and make a picture with some nice fog and the light hitting the trees I think the story changes alot and alot of people will see much more depth in the MF capture.
It's not ONLY the light, it's not ONLY the MF backs.
It's the combination of both.
But actually that's 101 photography
-
Given the number of people who got over 60%, and the fact that these were jpgs without direct comparisons I would say the evidence is clear that the difference CAN be seen. As a statistician I would say 60% plus is not going to be expected from random selections.,(I selected randomly, to get 44%, within is within standard error for a 100 N sample)
A lot of them looked like slr to me, not 3d, detailed, sharp without artifacts etc. that I expect from MFDB. Then I remembered a Steinway fellow saying he could make it sound like a Wurlitzer, or Knabe, or Baldwin. I don't think a Wurlitzer piano can be made to sound like a Steiway!
Good fun anyway, and some good shooting (making the point that you don't always need to top tool, as a MFDB (oops, maybe can't say that ! :-)
Victor
-
Given the number of people who got over 60%, and the fact that these were jpgs without direct comparisons I would say the evidence is clear that the difference CAN be seen. As a statistician I would say 60% plus is not going to be expected from random selections.,(I selected randomly, to get 44%, within is within standard error for a 100 N sample)
[a href=\"index.php?act=findpost&pid=138260\"][{POST_SNAPBACK}][/a]
It's way too early to draw any conclusions, certainly not "clear" evidence. I'm guessing that 60% success rate is within 1 standard deviation from 50% (ie. guessing). That would mean there is no statistically significant result of respondents accurately telling whether a shot is MDFB or dSLR. But I'll hold with statistical number crunching until I get a significant number of results - and we're not nearly there yet.
-
shame on me.
46%.
i am not worth to use mf gear.
i knew it ..
thanks a lot for the work you made yourself in preparing the test .....
-
61% correct.
Quentin
-
This is the most pointless and un-usefull test I have ever seen. Why don't you send us all 100 A4 prints and then we can judge properly.
JP Jespersen
http://www.ImagesOfNight.com (http://www.ImagesOfNight.com)
-
70%
But that is mostly because I was influenced by judging them on format, subject, and level of photography.
To put that survey together must have cost the better part of a day. Good initiative allthough
survey is majorly flawed in that it does not prove anything.
As was said before; we need large files of both slr and mfb in similar circumstances.
I can do a test with Nikon D2X and Hasselblad H3D, but I think it is a no-brainer that the mb would be better. Anyway this whole story started about Canon throwing the gauntlet.
So who are the mfdb and canon owners who can do it?
-
I got 41%. I must be blind. I guess the next questions is, How to get rid of my newly upgraded H3D-31 and go for a DSLR.
Captured with Hasselblad H1D (before the upgrade):
[attachment=3225:attachment]
-
51%, like "pps" (Paul). And like pps my choices were made on the same creterion: which format for that kind of job, since this seems to me the best criterion to be used here.
Impossible to make any other kind of selection, based on quality, "depth" (understand 3D), colours, sharpnes, DoF, etc ... of such images.
So does my and "pps"'s 51% suggest that 49% percent had chosen the wrong medium to shoot?
But that's another question.
Thierry
-
This is laughable. I am mad at myself for even posting on this thread.
Most people, even photographers, and myself, couldn't tell the difference of an 8X10 of a high end dslr or a MFDB exposed and processed properly, much less a web based jpeg! good gosh.
-
That's why I took in my opinion the only valid criterion to choose: based on what should be the logical format (for me) to take this shot.
All the rest is senseless, IMO.
As pointed out by someone else, I could put pics of my newly bought 10 MPx Lumix and nobody would see a difference!
Thierry
This is laughable. I am mad at myself for even posting on this thread.
Most people, even photographers, and myself, couldn't tell the difference of an 8X10 of a high end dslr or a MFDB exposed and processed properly, much less a web based jpeg! good gosh.
[a href=\"index.php?act=findpost&pid=138317\"][{POST_SNAPBACK}][/a]
-
That's why I took in my opinion the only valid criterion to choose: based on what should be the logical format (for me) to take this shot.
All the rest is senseless, IMO.
As pointed out by someone else, I could put pics of my newly bought 10 MPx Lumix and nobody would see a difference!
Thierry
[a href=\"index.php?act=findpost&pid=138319\"][{POST_SNAPBACK}][/a]
Oh no, how can you be using a tiny Lumix. Surely you know that 39MP is the minimum these days. You also need 16-bit otherwise the DR is so bad you cannot make out the freckles on your European skin under harsh Thai sunlight. I'm sure someone will have a few web-sized images to post just to confirm my point. Shame on you Thierry!
Henry
-
OK, I have a slightly less absurd, but still mostly stupid test. You can have a look at a few shoots I have up on my website (I so need to update with new stuff).
I've used 4 cameras in the Celeb section (loads automatically) Aptus 22 H1, P45 H1, Canon 5d, Canon 1ds II. For the most part my shots are F11+ and the crops are either double page or vertical making guessing based on crop a bit tougher.
Yes these are retouched some far more than others. Each shoot only uses 1 camera. So you at least have 1 photographers eye the same throughout the photos. Anybody think they can guess all 6 shoots right?
[url=http://www.jeffreyvogeding.com]My Webpage (http://www.jeffreyvogeding.com)
-
.... but I do only shoot Thai dark skins! So can I use my Lumix?
Thierry
Oh no, how can you be using a tiny Lumix. Surely you know that 39MP is the minimum these days. You also need 16-bit otherwise the DR is so bad you cannot make out the freckles on your European skin under harsh Thai sunlight. I'm sure someone will have a few web-sized images to post just to confirm my point. Shame on you Thierry!
Henry
[a href=\"index.php?act=findpost&pid=138327\"][{POST_SNAPBACK}][/a]
-
.... but I do only shoot Thai dark skins! So can I use my Lumix?
Thierry
[a href=\"index.php?act=findpost&pid=138331\"][{POST_SNAPBACK}][/a]
LOL
what do I know...I got 53% score so....
Henry
-
anyway better than me (51%)!
LOL
thierry
LOL
what do I know...I got 53% score so....
Henry
[a href=\"index.php?act=findpost&pid=138333\"][{POST_SNAPBACK}][/a]
-
Hey Thierry, I scored 52%! Can I have a Hy6 on loan now?
-Nik
anyway better than me (51%)!
LOL
thierry
[a href=\"index.php?act=findpost&pid=138337\"][{POST_SNAPBACK}][/a]
-
.... but I do only shoot Thai dark skins! So can I use my Lumix?
Thierry
[a href=\"index.php?act=findpost&pid=138331\"][{POST_SNAPBACK}][/a]
no you cannot, because the tiny sensor makes too much noise on brown skin. also we learned now that the 3d endering will be too less for too lo contrast from brown skin to green or brown backgrounds.
so it might happen that you will not see the model, at least not if it is nude ( what i will not hope anyway ).
-
first those having a lower score: they need to improve their skills!
Thierry
Hey Thierry, I scored 52%! Can I have a Hy6 on loan now?
-Nik
[a href=\"index.php?act=findpost&pid=138353\"][{POST_SNAPBACK}][/a]
-
again you have made your tests and learned your lesson: I am still in the trying phase.
thierry
no you cannot, because the tiny sensor makes too much noise on brown skin. also we learned now that the 3d endering will be too less for too lo contrast from brown skin to green or brown backgrounds.
so it might happen that you will not see the model, at least not if it is nude ( what i will not hope anyway ).
[a href=\"index.php?act=findpost&pid=138354\"][{POST_SNAPBACK}][/a]
-
OK, I have a slightly less absurd, but still mostly stupid test. You can have a look at a few shoots I have up on my website (I so need to update with new stuff).
I've used 4 cameras in the Celeb section (loads automatically) Aptus 22 H1, P45 H1, Canon 5d, Canon 1ds II. For the most part my shots are F11+ and the crops are either double page or vertical making guessing based on crop a bit tougher.
Yes these are retouched some far more than others. Each shoot only uses 1 camera. So you at least have 1 photographers eye the same throughout the photos. Anybody think they can guess all 6 shoots right?
[url=http://www.jeffreyvogeding.com]My Webpage (http://www.jeffreyvogeding.com)
[a href=\"index.php?act=findpost&pid=138329\"][{POST_SNAPBACK}][/a]
My guess :
#1 A22
#2 MkII
#3 5D
#4 5D
#5 MkII
#6 P45
Correct ??
-
As an emperor who wears only the very finest clothes that money can buy, I scored 100%, of course.
I'm surprised more so-called pros can't see the obvious (to emperors) 3D effect (is that the H3D effect or just an ordinary 3D effect ? Naturally, I can see both).
And those stunningly smooth transistions that even tiny, highly compressed jpgs can't hide. Not to mention color so real it bites yer bum off.
Really, there ought to be a forum just for emperors. Several here would qualify.
-
I got 54% bascially just judged from the feel of DOF. For web publishing it is rather difficult, if not impossible to judge. I suppose MFDB is not design primary for the web publishing (at common size) but rather than fine art and high quality print/output.
-
It's way too early to draw any conclusions, certainly not "clear" evidence. I'm guessing that 60% success rate is within 1 standard deviation from 50% (ie. guessing). That would mean there is no statistically significant result of respondents accurately telling whether a shot is MDFB or dSLR. But I'll hold with statistical number crunching until I get a significant number of results - and we're not nearly there yet.
[a href=\"index.php?act=findpost&pid=138265\"][{POST_SNAPBACK}][/a]
If the average goes above 60% then you are getting into 80-90% confidence that you can reject the null hypothesis (that is, that the choices are random) (SE is anot 10 out of 100, so, 45 to 55 is the expected score if the selections are basically 'random'))
-
Hehe, 57%. Well I guess That's $20k down the toilet then
I also couldn't find the "or film" option avalable on the nudes. They were so appalingly clichéd that I can only assume them to have been taken in a pre digital age.
-
This is the most pointless and un-usefull test I have ever seen. Why don't you send us all 100 A4 prints and then we can judge properly.
If you'd actually bothered to read the thread mentioned in the first post, you'd discover that some MFDB shooters were claiming that the difference between MFDB and DSLR was so great that the difference was obvious even in web JPEGs. Testing that claim is the point of the test.
Nobody is disputing that in studio conditions a MFDB is capable of producing a higher-quality file than a DSLR. What's being debated is whether that difference is detectable in web JPEGS.
-
Anybody else want to try my quiz before I provide results?
My guess :
#1 A22
#2 MkII
#3 5D
#4 5D
#5 MkII
#6 P45
Correct ??
[a href=\"index.php?act=findpost&pid=138359\"][{POST_SNAPBACK}][/a]
OK, I have a slightly less absurd, but still mostly stupid test. You can have a look at a few shoots I have up on my website (I so need to update with new stuff).
I've used 4 cameras in the Celeb section (loads automatically) Aptus 22 H1, P45 H1, Canon 5d, Canon 1ds II. For the most part my shots are F11+ and the crops are either double page or vertical making guessing based on crop a bit tougher.
Yes these are retouched some far more than others. Each shoot only uses 1 camera. So you at least have 1 photographers eye the same throughout the photos. Anybody think they can guess all 6 shoots right?
My Webpage (http://www.jeffreyvogeding.com)
-
Anybody else want to try my quiz before I provide results?
OK, I have a slightly less absurd, but still mostly stupid test. You can have a look at a few shoots I have up on my website (I so need to update with new stuff).
I've used 4 cameras in the Celeb section (loads automatically) Aptus 22 H1, P45 H1, Canon 5d, Canon 1ds II. For the most part my shots are F11+ and the crops are either double page or vertical making guessing based on crop a bit tougher.
Yes these are retouched some far more than others. Each shoot only uses 1 camera. So you at least have 1 photographers eye the same throughout the photos. Anybody think they can guess all 6 shoots right?
My Webpage
[a href=\"index.php?act=findpost&pid=138427\"][{POST_SNAPBACK}][/a] (http://www.jeffreyvogeding.com)
i let this pass without me.
´d like to see the results .....
-
It seems no one else wants to play. But our one contestant did very very well!!!!
His guesses
#1 A22
#2 MkII
#3 5D
#4 5D
#5 MkII
#6 P45
Actual
#1 A22
#2 P45
#3 5d
#4 5d
#5 MK II
#6 A22
HE got 4 out of 6!! And hes putting exact cameras with photos not just MF or DSLR!!!
IF we go MF/DLSR he got 5 out of 6!!! 83%%
That 67% correct wich i think is amazing, but I also think it shows you can tell!!!! I know I can. and its the reason why I chucked the 5D.
i let this pass without me.
´d like to see the results .....
[a href=\"index.php?act=findpost&pid=138471\"][{POST_SNAPBACK}][/a]
-
It seems no one else wants to play. But our one contestant did very very well!!!!
His guesses
#1 A22
#2 MkII
#3 5D
#4 5D
#5 MkII
#6 P45
Actual
#1 A22
#2 P45
#3 5d
#4 5d
#5 MK II
#6 A22
HE got 4 out of 6!! And hes putting exact cameras with photos not just MF or DSLR!!!
IF we go MF/DLSR he got 5 out of 6!!! 83%%
That 67% correct wich i think is amazing, but I also think it shows you can tell!!!! I know I can. and its the reason why I chucked the 5D.
[a href=\"index.php?act=findpost&pid=138474\"][{POST_SNAPBACK}][/a]
And here comes the fun part:
I have 20% eyesight on one eye and 60% on the other. Not a joke!
What would a not so blind photographer manage
Long live the MFDB!!!!
-
If you'd actually bothered to read the thread mentioned in the first post, you'd discover that some MFDB shooters were claiming that the difference between MFDB and DSLR was so great that the difference was obvious even in web JPEGs. Testing that claim is the point of the test.
Nobody is disputing that in studio conditions a MFDB is capable of producing a higher-quality file than a DSLR. What's being debated is whether that difference is detectable in web JPEGS.
[a href=\"index.php?act=findpost&pid=138415\"][{POST_SNAPBACK}][/a]
Sorry, I actually work and don't have all day to read through the threads. I can't beleive that anyone would think that web jpegs from any camera are going to be telling of the camera.
-
Sorry, I actually work and don't have all day to read through the threads. I can't beleive that anyone would think that web jpegs from any camera are going to be telling of the camera.
[a href=\"index.php?act=findpost&pid=138480\"][{POST_SNAPBACK}][/a]
Didnt I just prove you wrong?
-
Didnt I just prove you wrong?
[a href=\"index.php?act=findpost&pid=138486\"][{POST_SNAPBACK}][/a]
Not with a sample size of one (1).
-
Didnt I just prove you wrong?
[a href=\"index.php?act=findpost&pid=138486\"][{POST_SNAPBACK}][/a]
No you didnt, because anyone could have cheated on that test or been lucky. If they were all of the exact same subject matter, the test would be much more scientific. This is all ridiculous. I refuse to post anymore. We are running in circles.
I don't do work for web publishing, it is all for print, so this test is useless for my kind of work. It may apply to some people however.
-
No you didnt, because anyone could have cheated on that test or been lucky. If they were all of the exact same subject matter, the test would be much more scientific. This is all ridiculous. I refuse to post anymore. We are running in circles.
I don't do work for web publishing, it is all for print, so this test is useless for my kind of work. It may apply to some people however.
[a href=\"index.php?act=findpost&pid=138488\"][{POST_SNAPBACK}][/a]
First of all - I did not cheat.
Second - I may have been lucky, but I tried my best.
Third - I do not think people, who have made up their mind already,
will believe any outcome of any test that does not comply with what
they already "know" is the "truth".
I have a P30+ and a 5D.
I have had a 1Ds and a 1Ds Mk II.
I know it is a big difference on my screen between the P30+ and the 5D.
I know it is not that big on magazine paper.
And I do not care, because I love what I see on my screen and
how much more easy it is to work with P30+ files contra 5D files.
I feel happy when I look at a MFDB file and frustrated when I look at a DSLR file.
But thats only me.
Work with what you love/like and get you inspired to make
the very best pictures you can produce.
-
Ok, here are the results:
Total of 122 people responded (completed the quiz)
1/3 of respondents with 20+ years of photography experience
1/2 with 5-20 years
70% professionals
28% with 2+ years MFDB experience
100 shots, 50 MFDB and 50 dSLR
overall success rate: 56,52%. Please note that this means that with a strong (75+%) confidence we can not reject the possibility that respondents were guessing (50% success rate). In English: the respondents are no better at picking MFDB shots based on web-size JPEGs than by guessing.
Perhaps the most interesting part for most people follows.
These shots were overwhelmingly (75+%) voted to be dSLR shots, but were actually shot with a MFDB:
Shot #02 (http://harrijahkola.exposuremanager.com/scripts/expman.pl?rm=view_photo&photo_id=0253&file=02_medium.jpg), Shot #17 (http://harrijahkola.exposuremanager.com/scripts/expman.pl?rm=view_photo&photo_id=1717&file=17_medium.jpg&dir=galleries/19), #44 (http://harrijahkola.exposuremanager.com/scripts/expman.pl?rm=view_photo&photo_id=4446&file=44_medium.jpg&dir=galleries/19), #94 (http://harrijahkola.exposuremanager.com/scripts/expman.pl?rm=view_photo&photo_id=9435&file=94_medium.jpg&dir=galleries/19).
These shots were overwhelmingly (75+%) voted to be MFDB shots, but were actually shot with a dSLR:
#22 (http://harrijahkola.exposuremanager.com/scripts/expman.pl?rm=view_photo&photo_id=2245&file=22_medium.jpg&dir=galleries/19), #35 (http://harrijahkola.exposuremanager.com/scripts/expman.pl?rm=view_photo&photo_id=3554&file=35_medium.jpg&dir=galleries/19), #43 (http://harrijahkola.exposuremanager.com/scripts/expman.pl?rm=view_photo&photo_id=4341&file=43_medium.jpg&dir=galleries/19), #98 (http://harrijahkola.exposuremanager.com/scripts/expman.pl?rm=view_photo&photo_id=983&file=98_medium.jpg&dir=galleries/19).
And those who want all the details: Overall quiz report (http://app.sgizmo.com/reports/6094/7956/XODK6PQ791O8T06V8CQPOA754NNZ8D/) (WARNING: this page has each and every JPEG in one huge page, for a total of 15 megs or so)
These shots are MFDB:
1 2 8 9 13 15 16 17 20 24 25 26 27 29 30 32 33 34 37 38 40 44 47 48 51 53 54 56 59 62 63 64 65 67 70 71 75 77 78 79 82 84 85 89 90 94 95 99 100 101
These shots are dSLR:
3 4 5 6 7 10 11 12 14 18 19 21 22 23 28 31 35 36 39 41 42 43 45 46 49 50 52 55 57 58 60 61 66 68 69 72 73 74 76 80 81 83 86 87 91 92 93 96 97 98
To figure out the # of the shot, it's the 2 digits after photo_id in the full path, bolded below:
http://harrijahkola.exposuremanager.com/sc...photo&photo_id= (http://harrijahkola.exposuremanager.com/scripts/expman.pl?rm=view_photo&photo_id=)0823&file=08_medium.jpg&dir=galleries/19
Unfortunately it appears that SurveyGizmo doesn't allow for any cross-referencing of data. It doesn't even give out results by responder, only by question #. I even had to do all the above calcs manually, although it's supposed to support quizzes. I guess you get what you pay for: the service is free.
-
These shots are MFDB:
1 2 8 9 13 15 16 17 20 24 25 26 27 29 30 32 33 34 37 38 40 44 47 48 51 53 54 56 59 62 63 64 65 67 70 71 75 77 78 79 82 84 85 89 90 94 95 99 100 101
[a href=\"index.php?act=findpost&pid=138675\"][{POST_SNAPBACK}][/a]
Are you sure these are all shot with digital backs? Several look very much like film...
-
Are you sure these are all shot with digital backs? Several look very much like film...
[a href=\"index.php?act=findpost&pid=138694\"][{POST_SNAPBACK}][/a]
Somebody else can set up the quiz for film vs digital debate
There could very well be film shots. I relied entirely on what the photographer reported. If they said it was digital, it's digital.