Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: evgeny on May 31, 2009, 06:48:16 pm
-
Hi,
I had fun last night. This is a not very accurate test, but it shows something. Images are slightly sharpened in Photoshop (Adaptive Sharpening % in Sharpener Pro plugin). The real color is produced by the multi-shot back
Nikon D40 with 50mm f1.4, f16, ISO 200 (min), 5942K, CS4 sharp 40%
(http://forum.selldesk.com/forum/upload/images/Evgeny/studio/2_Nikon_f16_ISO200_5942K_sharp40_900px.jpg)
Contax 645 with 140mm, Aptus 65, f13, ISO 50, CS4 sharp 24%, mirror up
(http://forum.selldesk.com/forum/upload/images/Evgeny/studio/3_Aptus_f13_ISO50_sharp24_mirror_up_900px.jpg)
Contax 645 with 140mm, Aptus 65, f13, ISO 50, CS4 sharp 24%, mirror up, Grey Card
(http://forum.selldesk.com/forum/upload/images/Evgeny/studio/3_Aptus_f13_ISO50_sharp24_mirror_up_wb_900px.jpg)
Contax 645 with 140mm, Aptus 65, f9.5, ISO 50, CS4 sharp 24%, mirror up
(http://forum.selldesk.com/forum/upload/images/Evgeny/studio/3_Aptus_f9.5_ISO50_sharp24_mirror_up_900px.jpg)
Contax 645 with 140mm, Aptus 65, f9.5, ISO 50, CS4 sharp 24%, mirror up, Grey Card
(http://forum.selldesk.com/forum/upload/images/Evgeny/studio/3_Aptus_f9.5_ISO50_sharp24_mirror_up_wb_900px.jpg)
Contax 645 with 120mm, Sinar 54H, f9.5, ISO 40, 1 shoot, CS4 sharp 24%, shutter triggered from Macbook
(http://forum.selldesk.com/forum/upload/images/Evgeny/studio/4_Sinar54_f9.5_ISO40_1shoot_sharp24_900px.jpg)
Contax 645 with 120mm, Sinar 54H, f9.5, ISO 40, 4 shoot (4 seconds delay), CS4 sharp 24%, shutter triggered from Macbook
(http://forum.selldesk.com/forum/upload/images/Evgeny/studio/5_Sinar54_f9.5_ISO40_4shoot_sharp24_900px.jpg)
Contax 645 with 120mm, Sinar 54H, f9.5, ISO 40, 16 shoot (4 seconds delay), CS4 sharp 48%, shutter triggered from Macbook
(http://forum.selldesk.com/forum/upload/images/Evgeny/studio/6_Sinar54_f9.5_ISO40_16shoot_sharp48_900px.jpg)
Contax 645 with 120mm, Sinar 54H, f9.5, ISO 40, 1 shoot, CS4 sharp 24%, shutter triggered from Macbook, Grey Card
(http://forum.selldesk.com/forum/upload/images/Evgeny/studio/4_Sinar54_f9.5_ISO40_1shoot_sharp24_wb_900px.jpg)
-
As expected the colors look like crap from the Nikon...:+}
Snook
-
As expected the colors look like crap from the Nikon...:+}
Snook
Hi Snook,
Well it just goes to show personal preference. I think the Nikon has the nicest colour.
Cheers
Simon
-
I personally think its hard to tell since the exposures are different. The Nikon looks much brighter then the Leaf Files. I think I would want to see the RGB numbers of the raw files, and then process them.
-
I personally think its hard to tell since the exposures are different. The Nikon looks much brighter then the Leaf Files. I think I would want to see the RGB numbers of the raw files, and then process them.
Agree. There is no shadow detail in the right doll's leg.
Exposure different make a lot of color differences.
-
I personally think its hard to tell since the exposures are different. The Nikon looks much brighter then the Leaf Files. I think I would want to see the RGB numbers of the raw files, and then process them.
Hi.
You can download the original 5Mb Nikon RAW (http://forum.selldesk.com/forum/upload/images/Evgeny/studio/tmp/2_Nikon_f16_ISO200.NEF) file.
Please note, the color temperature of the original is 5000K. The image I posted here has color temperature 5942K to match the Leaf temperature.
-
Wouldn't the advantages of multishot be totally lost by the time you reduce the file to this size? (except for moire reduction which doesn't appear to be an issue here)
-
Wouldn't the advantages of multishot be totally lost by the time you reduce the file to this size? (except for moire reduction which doesn't appear to be an issue here)
Graham, the size does matter.
-
Are these links still good? I just get missing-image-icons.
-
Same here.
Thierry
Are these links still good? I just get missing-image-icons.
-
Hi
I just renewed my internet domain, it expired yesterday. The images should be back
-
Agree. There is no shadow detail in the right doll's leg.
Exposure different make a lot of color differences.
The minimum ISO in Nikon D40 is 200.
So, I set f16 to use the same shutter speed and light.
-
There is a noticeable difference between Aptus and 54H in red color, look at the left doll "t-shirt".
Update: I set camera with Aptus to f13, while camera with 54H was set to f9.5. The reason is to get "128" value in the Aptus gray card, which indicates proper exposure.
-
There is a noticeable difference between Aptus and 54H in red color, look at the left doll "t-shirt"
As others have pointed out, the Nikon shot is too bright in comparison to make any conclusions, even about colour, but anyway, comparing a back to an ancient 6MP camera is not that useful. What I find interesting is the large colour difference between the Leaf and Sinar shots, with the Sinar colour being much closer to the Nikon colour. The reds on the Leaf are very muted, almost dark, in comparison, whereas the oranges are almost untouched. Weird. And the yellow is greenish in places in the Leaf file.
If you are in the mood, could you make a new comparison just between the Leaf and 1-shot 54H, with a brighter exposure? Larger images would also be interesting, if you can host them.
-
As others have pointed out, the Nikon shot is too bright in comparison to make any conclusions, even about colour, but anyway, comparing a back to an ancient 6MP camera is not that useful. What I find interesting is the large colour difference between the Leaf and Sinar shots, with the Sinar colour being much closer to the Nikon colour. The reds on the Leaf are very muted, almost dark, in comparison, whereas the oranges are almost untouched. Weird. And the yellow is greenish in places in the Leaf file.
If you are in the mood, could you make a new comparison just between the Leaf and 1-shot 54H, with a brighter exposure? Larger images would also be interesting, if you can host them.
I hope to find the time to do more tests.
Please suggest how to better test these sensors, such as, which better subject that I can easily find, ISO, f-stops, exposures, light, etc.
-
I hope to find the time to do more tests.
Please suggest how to better test these sensors, such as, which better subject that I can easily find, ISO, f-stops, exposures, light, etc.
Actually, it is a nice test, IMO. Perhaps a colour checker next to the dolls, and a separate shot of someone to see the skin colours.
-
There is a noticeable difference between Aptus and 54H in red color, look at the left doll "t-shirt".
The 54H has a tendency to oversaturate reds when using the default profile in Captureshop.
-
Hi Snook,
Well it just goes to show personal preference. I think the Nikon has the nicest colour.
Cheers
Simon
I don't see how anyone can judge what color is the nicest or the most accurate without seeing the original subject. Maybe the Nikon's colors are the "nicest" but that is subjective and has no bearing on accuracy. Perhaps a grey card or step chart should have been included so that exposure differences could have been compensated for, and there would be a consistent base white balance. Even then, the OP would have to tell us which shot produces the most accurate colors in relation to the physical, real, doll.
-
the OP would have to tell us which shot produces the most accurate colors in relation to the physical, real, doll.
Hi, what is OP?
I will include grey card in the image next time.
I use grey card, don't have colour checker. Do I really need colour checker?
-
Hi, what is OP?
I will include grey card in the image next time.
I use grey card, don't have colour checker. Do I really need colour checker?
OP = Original Poster
I don't know what you really "need." I'm just making the point that the rest of us simply cannot judge on color accuracy without a "reference". Therefore, some sort of "known" standard like a color checker or grey card would be a start, but ultimately, it's probably your own judgement, since you're the only one who can see the original scene and the captures simultaneously.
-J
-
I just added three images for Aptus:
f13 with Grey Card
f9.5
f9.5 with Grey Card
-
I just added three images for Aptus:
f13 with Grey Card
f9.5
f9.5 with Grey Card
Well the grey card certainly shows the different exposures, but also, having had a Leaf back and using Leaf Capture, I know that the colour
you get will depend on what colour profile you have chosen... ie. for Product, Portrait, stronger or weaker yellows etc.
But a good start for a test would be to make sure each image including the Nikon was exposed to give the same reading in the Grey card.
But thanks for your time and interest anyhow.
ta
Andrew
-
I know that the colour you get will depend on what colour profile you have chosen... ie. for Product
I used Aptus 65 Product.
I just also added an image with Grey Card for 54H.
-
You are busting your butt to provide information for everyone here.
Irregardless of your testing method, thank you for taking the time.
Mark
-
Hi Evgeny,
Thanks for posting these tests. I also find that the Sinar 54H currently gives the highest quality digital capture from any MFDB for use in the studio, due to its microscanning capability. It is no surprise though, since it can physically capture a whopping 16 x 22MP = 352MP of image information and produces a final 264MP image. This is more than 4x the amount of pixels than the highest resolution single shot backs can capture, such as the P65+.
If you want to have more fun, you can put the 54H on a Sinar P2 with a Macroscan unit attached to it. The macroscan unit will work with the 54H, and will automatically move the 54H up and down across an approximate 58mm x 72mm area to capture four shots and automatically stitch them into a single image. However, each individual shot can also be a 16-shot microscan, so you will end up with a final image of a little more than 1000 megapixels of physically captured information (no interpolation). This kind of technique is being used by museums and other institutes that are trying to preserve features of ancient relics and artwork.
A while back Sinar posted an article highlighting this technique, which you can find here: http://www.sinar.ch/file_uploads/bibliothe...umofbeijing.pdf (http://www.sinar.ch/file_uploads/bibliothek/k_125_SinarCaseStories/638_0_casestory_repro_capitalmuseumofbeijing.pdf)
David
-
Wonderful memories, since I had the unique opportunity and chance to enter the highly secretive and well guarded places of these 2 museums and see some unique pieces of paintings old of a few centuries, while training their photographers with the system. Those paintings are only reproduced and never shown in public.
The results of the 16-shot reproductions can be seen in a special room of the museum, on a few 20" Cinema Displays and a mouse to zoom in and see the details captured: amazing!
Thierry
http://www.sinar.ch/file_uploads/bibliothe...umofbeijing.pdf (http://www.sinar.ch/file_uploads/bibliothek/k_125_SinarCaseStories/638_0_casestory_repro_capitalmuseumofbeijing.pdf)
David
-
David, I thought that 4-shot yielded a 22MP image with full colour sampling, and 16-shot an 88MP image with full colour sampling? There are 352 million pixels being sampled, but by the time the image is ready, there are only 88MP, right? Which is still a lot, and I guess these images upscale brilliantly, in case more is needed.
C
Hi Evgeny,
Thanks for posting these tests. I also find that the Sinar 54H currently gives the highest quality digital capture from any MFDB for use in the studio, due to its microscanning capability. It is no surprise though, since it can physically capture a whopping 16 x 22MP = 352MP of image information and produces a final 264MP image. This is more than 4x the amount of pixels than the highest resolution single shot backs can capture, such as the P65+.
If you want to have more fun, you can put the 54H on a Sinar P2 with a Macroscan unit attached to it. The macroscan unit will work with the 54H, and will automatically move the 54H up and down across an approximate 58mm x 72mm area to capture four shots and automatically stitch them into a single image. However, each individual shot can also be a 16-shot microscan, so you will end up with a final image of a little more than 1000 megapixels of physically captured information (no interpolation). This kind of technique is being used by museums and other institutes that are trying to preserve features of ancient relics and artwork.
A while back Sinar posted an article highlighting this technique, which you can find here: http://www.sinar.ch/file_uploads/bibliothe...umofbeijing.pdf (http://www.sinar.ch/file_uploads/bibliothek/k_125_SinarCaseStories/638_0_casestory_repro_capitalmuseumofbeijing.pdf)
David
-
That's correct, Carsten. I believe David was refering to the number of pixels capturing information, when the 16-shot mode is used, and this is effectively 352 MP. When sampling is done, it needs the information from 4 pixels (1 Red, 2 Green, 1 Blue) to get the full un-interpolated color information of each single 88 M pixels.
Thierry
David, I thought that 4-shot yielded a 22MP image with full colour sampling, and 16-shot an 88MP image with full colour sampling? There are 352 million pixels being sampled, but by the time the image is ready, there are only 88MP, right? Which is still a lot, and I guess these images upscale brilliantly, in case more is needed.
C
-
Hi David,
thank you for the 54H, it's superb. I also sold my 54M yesterday, so all deals went so well!
I will shoot with the 54H back at my still life workshop that I will teach by the end of this months.
The multi-shot was easy with my Contax 645. I mount the camera on Neotec tripod.
The 4-shot images were sharp just out of the camera.
The 16-shot images need (and allow?) more sharpening in photoshop. I think how to fix the mirrow for the entire capture to make 16-shot as sharp as 4-shot.
The 16-shot images show darker pixels on some corners, which look like natural shadows, not shifts between captures. These darker pixel not shown in 1 and 4-shot images at 100% view. Is that a better dynamic range or what?
I don't have a view camera, yet, and will need more a power computer to proceed such very big images
I heard microscan adapter has some limitations, does it work with the new CaptureShop 5.6.3?
I think a 88 megapixels image can be printed on approximately 3' (1 meter) page at 300 dpi. That's big and should show every details at close distance!
-
The 16-shot images need (and allow?) more sharpening in photoshop.
Yes, the 16-shot definitively needs and allows for more sharpening.
I think how to fix the mirrow for the entire capture to make 16-shot as sharp as 4-shot.
Rainer, e.g., has used the 16-shot on a Contax and has dampered the mirror movement with some special foam: I believe he can say more here.
The 16-shot images show darker pixels on some corners, which look like natural shadows, not shifts between captures. These darker pixel not shown in 1 and 4-shot images at 100% view. Is that a better dynamic range or what?
I have no explanation for that, and am not sure to understand what it is.
I heard microscan adapter has some limitations, does it work with the new CaptureShop 5.6.3?
Captureshop 4.x.x is the latest version of Captureshop supporting the Macroscan feature.
I think a 88 megapixels image can be printed on approximately 3' (1 meter) page at 300 dpi. That's big and should show every details at close distance!
Withouh any problem, at least!
Best regards,
Thierry
-
Thierry, thank you!
-
This is a 100% view of the 4-shoot image. It's better to see it at 200% to compare to the 16-shoot image below
(http://forum.selldesk.com/forum/upload/images/Evgeny/studio/tmp/6_Sinar54_f9.5_ISO40_4shoot_nodarkpixels.jpg)
This is a 100% view of the 16-shoot unprocessed image. Please see darker pixels on the right side of the doll. There is no such pixels on the left side.
I don't know what is it. It not looks like a shift during 16 exposures.. Comments?
(http://forum.selldesk.com/forum/upload/images/Evgeny/studio/tmp/7_Sinar54_f9.5_ISO40_16shoot_darkpixels.jpg)
-
Dear Evgeny,
I see what you mean, now: an image is worth a 1000 words!
These what you call "black pixels" simply indicates a pattern following a misregistration due to some movement or vibrations during the process of the 16 shot.
The 16 shot mode is critical: one has to have a very stable floor, a heavy and stable stand (tripod not recommended), all machines likely to produce some sort of vibrations in the room or nearby rooms should be shut off (e.g. AC), walking during the process should be avoided, etc ...
I can tell you that I twice had such a problem with AC elements on the ceiling, and the other on the floor, about 15 meters aways from the shooting place. I had as well problems with thiny vibrations on a groundfloor although having a very stable concrete floor.
Therefore, I am convinced that vibrations are your problem here.
Best regards,
Thierry
This is a 100% view of the 4-shoot image. It's better to see it at 200% to compare to the 16-shoot image below
This is a 100% view of the 16-shoot unprocessed image. Please see darker pixels on the right side of the doll. There is no such pixels on the left side.
I don't know what is it. It not looks like a shift during 16 exposures.. Comments?
-
These what you call "black pixels" simply indicates a pattern following a misregistration due to some movement or vibrations during the process of the 16 shot.
Hi Thierry,
It's good to know what's exactly the problem
I don't have a Foba stand, but will try to add a weight to the base of tripod, and will try a wooden Berlebach tripod, which can support heavy loads.
Thanks
Evgeny
-
That's correct, Carsten. I believe David was refering to the number of pixels capturing information, when the 16-shot mode is used, and this is effectively 352 MP. When sampling is done, it needs the information from 4 pixels (1 Red, 2 Green, 1 Blue) to get the full un-interpolated color information of each single 88 M pixels.
Thierry
I think this is an important point to distinguish. The 16-shot mode of the 54H is really much much more than just an 88MP image, since the amount of physically captured information is four colors at each pixel site. A 60MP single-shot back only captures one color at each pixel location for a total of 60MP, whereas a 88MP multi-shot back captures a total of 16x22MP=352MP. A back such as the P65+ captures one-third of the RGB image information (60MP), and needs to estimate the missing two-thirds (120MP) of color information to produce a final RGB image of 180MP (R,G,B at every pixel site). The multi-shot back does not have any estimated pixel information.
However, if we do allow the same degree of pixel estimation from a 54H file and allow its physically captured information to be up-sized with two-thirds of pixel estimation, you would get a final RGB image of size 264MP x 3 = 792MP. So, the resulting RGB file in comparison to a single shot back like the P65+ is more than 4X (ie, 792 / 180). This would be the more accurate comparison to a single shot back. The extreme image detail from the 54H 16-shot mode has been confirmed in practice, especially by the museum / preservation photographers as I noted above.
David
-
I think a 88 megapixels image can be printed on approximately 3' (1 meter) page at 300 dpi. That's big and should show every details at close distance!
Yes, this is true. If you want to be able to visually see print details of 300 dpi, then the 54H file can be printed as large as about 0.75m x 1m. This assumes that your image actually contains such tiny details. This kind of resolution at this size of print implies a captured resolution of over 110 lp/mm at the sensor. So, if your image does not actually have this kind of detail, then of course you can print much larger.