FAQ FAQ  Forum Search   Events   Register Register  Login Login

The joy of 1.54m pixels.

Page  123 5>
Author
Jonas A-R View Drop Down
Senior Member
Senior Member

Joined: 29 December 2007
Country: Denmark
Location: Denmark
Status: Online
Posts: 1467
Post Options Post Options   Quote Jonas A-R Quote  Post ReplyReply Direct Link To This Post Topic: The joy of 1.54m pixels.
    Posted: 31 August 2014 at 14:54
It is often discussed how 36MP on a full frame sensor is a challenge for our lenses. The perceptual MP score from DxOmark seems to suggest that the lenses are limiting since they resolve less than a perfect lens would on a sensor with the nominal MP count under test. I think this is rubbish.

I have recently purchased a Pentax Q camera with a tiny 12MP sensor. If a FF sensor was build to the same specs, it would have 360MP. I also purchased a cheap ebay adapter allowing me to study the performance of alpha mount lenses in greater detail than previously.

Today I will post the very first test result. If people are interested, feel free to suggest tests of interest, and I will try to see what I can do as time permits.
I will definitely try to document aliasing which is proof of undersampling. I will also test a range of lenses from cheap to expensive and perhaps I will even try to see how these tiny pixels perform in low light exposures   

This is the setup

Sony 135/1.8 on the tiny Q

This is the view of the test subject as seen from my a900 (the distance is a little more than 11m)


And now to some picture samples. The adapter does not state the aperture. It has a stepless ring to control it. I therefore set the camera up manually, took a reading fully open and then stopped down until the meter suggested a shutterspeed 4x as fast (2 stops = ~f/3.6). I used this aperture because a 350MP Bayer sensor is just about sufficient to be able to record diffraction in all color channels at f/4.

First the result from the Q (100% crop)


Then the result from the a900 at f/4


I think such data is sufficient to prove that we have some way to go before we are even close to proper sampling of the analog signal projected by our lenses. I also conclude that the 135/1.8 is pretty sharp

Edited by Jonas A-R - 31 August 2014 at 15:38
a9 a6300 21/2.8 Loxia 35/2.8Z 35/1.4Z 50/1.4Z 55/1.8Z 85/1.4GM 90/2.8G 12-24/4G 24-105/4G 70-200/4G 100-400/4-5.6GM 2x TC
 



Back to Top
pegelli View Drop Down
Admin Group
Admin Group
Dyxum Administrator

Joined: 02 June 2007
Country: Belgium
Location: Schilde
Status: Offline
Posts: 29292
Post Options Post Options   Quote pegelli Quote  Post ReplyReply Direct Link To This Post Posted: 31 August 2014 at 15:24
Fantastic test Jonas, I think we can learn a lot that way.

What's the crop factor of the Q?

Also, it seems the lens mount and adapter is pretty strong if you can just hang the camera on the tripod and let all that bulk of the giant 135/1.8 in front just hang free. I think you might be better off in the long run if you can support the lens better.
Mind the bandwidth of others, don't link pictures larger then 1024 wide or 960 pix high, see here
Back to Top
Jonas A-R View Drop Down
Senior Member
Senior Member

Joined: 29 December 2007
Country: Denmark
Location: Denmark
Status: Online
Posts: 1467
Post Options Post Options   Quote Jonas A-R Quote  Post ReplyReply Direct Link To This Post Posted: 31 August 2014 at 15:45
Originally posted by pegelli pegelli wrote:

Fantastic test Jonas, I think we can learn a lot that way.


Thanks, and I do hope that we may learn something

Originally posted by pegelli pegelli wrote:

What's the crop factor of the Q?

It has a sensor diagonal of 7.7mm so that would be 5.6. It has a 4:3 aspect ratio so I suppose it depends on the aspect ratio of the displayed photo

Originally posted by pegelli pegelli wrote:

Also, it seems the lens mount and adapter is pretty strong if you can just hang the camera on the tripod and let all that bulk of the giant 135/1.8 in front just hang free. I think you might be better off in the long run if you can support the lens better.


Yes, I agree. It looks dangerous. I just had to test my sharpest lens first :) Do you know of any cheap support system that I could get for this? The alpha lenses protudes way beyond the base of the camera, so I think I need some sort of spacer?

a9 a6300 21/2.8 Loxia 35/2.8Z 35/1.4Z 50/1.4Z 55/1.8Z 85/1.4GM 90/2.8G 12-24/4G 24-105/4G 70-200/4G 100-400/4-5.6GM 2x TC
Back to Top
pegelli View Drop Down
Admin Group
Admin Group
Dyxum Administrator

Joined: 02 June 2007
Country: Belgium
Location: Schilde
Status: Offline
Posts: 29292
Post Options Post Options   Quote pegelli Quote  Post ReplyReply Direct Link To This Post Posted: 31 August 2014 at 15:55
Jonas, how about something like this

Don't know if it has enough flexibility in the vertical direction, but if not you can try find some spacer to go under the camera in addition.

You can also try to produce something similar as a DIY project.

Edited by pegelli - 31 August 2014 at 15:59
Mind the bandwidth of others, don't link pictures larger then 1024 wide or 960 pix high, see here
Back to Top
Jonas A-R View Drop Down
Senior Member
Senior Member

Joined: 29 December 2007
Country: Denmark
Location: Denmark
Status: Online
Posts: 1467
Post Options Post Options   Quote Jonas A-R Quote  Post ReplyReply Direct Link To This Post Posted: 31 August 2014 at 16:08
Thanks! I think I may have to go for some DIY due to the price but that thing should be a good inspiration for a litlle project :)
a9 a6300 21/2.8 Loxia 35/2.8Z 35/1.4Z 50/1.4Z 55/1.8Z 85/1.4GM 90/2.8G 12-24/4G 24-105/4G 70-200/4G 100-400/4-5.6GM 2x TC
Back to Top
Jonas A-R View Drop Down
Senior Member
Senior Member

Joined: 29 December 2007
Country: Denmark
Location: Denmark
Status: Online
Posts: 1467
Post Options Post Options   Quote Jonas A-R Quote  Post ReplyReply Direct Link To This Post Posted: 31 August 2014 at 20:21
It seems I got some images mixed up. It's a pain when there is no exif telling the aperture :)
The image posted above showed the lens a ~f/2.5, one stop down.

Here is a series showing f/1.8, f/2.5, f/3.6 & f/5 (approximate f-numbers as described)

f/1.8

f/2.5

f/3.6

f/5


That lens impresses me
a9 a6300 21/2.8 Loxia 35/2.8Z 35/1.4Z 50/1.4Z 55/1.8Z 85/1.4GM 90/2.8G 12-24/4G 24-105/4G 70-200/4G 100-400/4-5.6GM 2x TC
 



Back to Top
Blame View Drop Down
Senior Member
Senior Member

Joined: 14 October 2010
Country: United Kingdom
Location: London UK
Status: Offline
Posts: 2626
Post Options Post Options   Quote Blame Quote  Post ReplyReply Direct Link To This Post Posted: 31 August 2014 at 21:04
Um.

From what I have read such as the Nikon D810 have no problem delivering full resolution to the point of aliasing providing:

a) A quality prime.
b) A limited range of apertures.
c) You ignore the sides and corners.

I dare say a FF camera with 1.5um pixels would deliver far more center resolution but the corners are really really going to look silly.
A900, Min 24-105, 35-105, Samyang 14/2.8, 35/1.4, Sig 70/2.8 Macro, ISCO Ultra 125/2, Tam 180/3.5 Macro, Sig 400/5.6 TeleMacro
Back to Top
Jonas A-R View Drop Down
Senior Member
Senior Member

Joined: 29 December 2007
Country: Denmark
Location: Denmark
Status: Online
Posts: 1467
Post Options Post Options   Quote Jonas A-R Quote  Post ReplyReply Direct Link To This Post Posted: 31 August 2014 at 21:42
Originally posted by Blame Blame wrote:

I dare say a FF camera with 1.5um pixels would deliver far more center resolution but the corners are really really going to look silly.


Unfortunately I cannot show that. But of course you are right: Lenses perform best in the center.
a9 a6300 21/2.8 Loxia 35/2.8Z 35/1.4Z 50/1.4Z 55/1.8Z 85/1.4GM 90/2.8G 12-24/4G 24-105/4G 70-200/4G 100-400/4-5.6GM 2x TC
Back to Top
luke View Drop Down
Senior Member
Senior Member

Joined: 09 April 2013
Status: Offline
Posts: 892
Post Options Post Options   Quote luke Quote  Post ReplyReply Direct Link To This Post Posted: 31 August 2014 at 23:07
Nice test. TFS. Now you only need that shift adapter....
Back to Top
kerrath View Drop Down
Senior Member
Senior Member

Joined: 07 July 2012
Country: United States
Location: LowerCalifornia
Status: Offline
Posts: 1672
Post Options Post Options   Quote kerrath Quote  Post ReplyReply Direct Link To This Post Posted: 31 August 2014 at 23:18
I am impressed.
I would never have thought even a good zeiss prime would perform that well on such tiny pixels on a tiny sensor.

I would argue, however, that rather than the lens failing to deliver as pixels grow smaller, the user would. Increasingly small amounts of shake would obscure the details of the image, causing a denser sensor to only render gains when used on a tripod. In that, I feel like the current pixel density of sensors is on par with our ability to hold cameras steady, and the ability of lenses/cameras to compensate where we fall short.
Back to Top
thornburg View Drop Down
Senior Member
Senior Member

Joined: 25 July 2013
Country: United States
Location: PA
Status: Offline
Posts: 3765
Post Options Post Options   Quote thornburg Quote  Post ReplyReply Direct Link To This Post Posted: 01 September 2014 at 03:14
Originally posted by kerrath kerrath wrote:

I am impressed.
I would never have thought even a good zeiss prime would perform that well on such tiny pixels on a tiny sensor.

I would argue, however, that rather than the lens failing to deliver as pixels grow smaller, the user would. Increasingly small amounts of shake would obscure the details of the image, causing a denser sensor to only render gains when used on a tripod. In that, I feel like the current pixel density of sensors is on par with our ability to hold cameras steady, and the ability of lenses/cameras to compensate where we fall short.


Well, it would make sense to me if you applied the crop factor as part of the minimum shutter speed rule of thumb, which would make the minimum shutter for the Zeiss 135/1.8 used on the Q 1/800th. (EDIT: That's for hand-held use, of course).

In other words, even a 50mm lens on that camera is a long telephoto, requiring a steady hand and/or a stable support.
Sony a3000, a6000, a57, a99 - Sony E 16-50, 28/2 | Vivitar 13, 85 | Minolta 24, 28-105, 35-105, 50/1.7, 75-300 | Tokina 28-70/2.6-2.8 | Sigma 70/2.8 Macro | Tamron 70-200/2.8 | Celestron 1000/11
Back to Top
Jonas A-R View Drop Down
Senior Member
Senior Member

Joined: 29 December 2007
Country: Denmark
Location: Denmark
Status: Online
Posts: 1467
Post Options Post Options   Quote Jonas A-R Quote  Post ReplyReply Direct Link To This Post Posted: 01 September 2014 at 05:46
Originally posted by kerrath kerrath wrote:

I am impressed.
I would never have thought even a good zeiss prime would perform that well on such tiny pixels on a tiny sensor.

I would argue, however, that rather than the lens failing to deliver as pixels grow smaller, the user would. Increasingly small amounts of shake would obscure the details of the image, causing a denser sensor to only render gains when used on a tripod. In that, I feel like the current pixel density of sensors is on par with our ability to hold cameras steady, and the ability of lenses/cameras to compensate where we fall short.


It is true that the return will diminsh as pixels shrink and will require good technique. However, a sharp lens and good technique on one of the modern AA filter free cameras will provide an image which is full of alising. Most people just don't see it unless it is obvious such as Moir. The false details can easily be mistaken as sharpness.

a9 a6300 21/2.8 Loxia 35/2.8Z 35/1.4Z 50/1.4Z 55/1.8Z 85/1.4GM 90/2.8G 12-24/4G 24-105/4G 70-200/4G 100-400/4-5.6GM 2x TC
Back to Top
kerrath View Drop Down
Senior Member
Senior Member

Joined: 07 July 2012
Country: United States
Location: LowerCalifornia
Status: Offline
Posts: 1672
Post Options Post Options   Quote kerrath Quote  Post ReplyReply Direct Link To This Post Posted: 01 September 2014 at 12:18
Originally posted by thornburg thornburg wrote:

Well, it would make sense to me if you applied the crop factor as part of the minimum shutter speed rule of thumb, which would make the minimum shutter for the Zeiss 135/1.8 used on the Q 1/800th. (EDIT: That's for hand-held use, of course).

In other words, even a 50mm lens on that camera is a long telephoto, requiring a steady hand and/or a stable support.
That's a way to think about it, but rather than going for the crop factor of the Q, I prefer to think of a crop factor of the individual pixels since I am conceptually comparing a future high density FF sensor to a modern FF sensor.

It should be a crop factor of roughly 3:1 (2.93) on a pixel to pixel basis. So shake causing 1 pixel of blur on the 36mp sensor would cause 3 pixels of blur on the high density sensor. Thus necessitating faster exposures or significantly improved user ability, or significantly improved mechanical shake reduction as part of the hardware.

So a minimum shutter speed of 1/400th for a FF sensor with the same pixel density as the Q, or, 3 stops of shake reduction, to produce the absolute amount of shake-blur, thereby not negating the potential gains by taking advantage of the lens' best ability to resolve an image.

The same would hold true for the Q, in terms of absolute resolution. For conventional cropping by sensor size, your estimate is still correct.
Back to Top
revdocjim View Drop Down
Moderator Group
Moderator Group

Joined: 11 September 2006
Country: Japan
Location: Mt. Akagi
Status: Offline
Posts: 8608
Post Options Post Options   Quote revdocjim Quote  Post ReplyReply Direct Link To This Post Posted: 01 September 2014 at 12:29
OK, I confess to being completely confused by all of this. Why is the A900 image in the OP so badly pixelated and what exactly is being proven with all of this? Sorry, I'm just not very quick with the tech stuff.
Gallery A7S, A7Rii, Batis 18/2.8, 25/2 Sony 35/2.8, 55/1.8, 90/2.8M, 24-105/4, Minolta 135STF, 200/2.8 Blog
Back to Top
Dyxum main page >  Forum Home > Equipment forums > Camera Talk Page  123 5>

Forum Jump Forum Permissions View Drop Down



This page was generated in 0.078 seconds.

Monitor calibration strip

Dyxum.com - Home of the alpha system photographer

In memory of Cameron Hill - brettania

Feel free to contact us if needed.