Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
04-25-2014, 01:55 AM   #1
Veteran Member
zoolander's Avatar

Join Date: Dec 2012
Location: Brisbane
Photos: Albums
Posts: 334
K-5 anti-aliasing filter sharpness

I'm having an issue. I did a little sharpness test with my 18-55mm and Pentax FA 50mm f/1.4. I didn't see a lot of difference between two lenses at what I believe is their sharpest F-stop (5.6). The FA 50 1.4 tested a lot sharper on Photozone compared to the 18-135.

I remember reading somewhere that the K-5 can inhibit sharp lenses due to the anti-aliasing filter being quite strong.

So perhaps even buying more expensive lenses there'll be a "glass ceiling" (pardon the pun) to sharpness, because of the AA filter ......

Is this the case ? Or am I in error ?

04-25-2014, 01:59 AM   #2
Administrator
Site Webmaster
Adam's Avatar

Join Date: Sep 2006
Location: Arizona
Photos: Gallery | Albums
Posts: 40,958
QuoteOriginally posted by zoolander Quote
Is this the case ? Or am I in error ?
You should easily be able to tell the difference between the two lenses- were you on a tripod when doing the tests? The whole AA filter thing was blown way out of proportion when filterless cameras came out and it really doesn't play a big role compared to resolution and dynamic range.

Adam
PentaxForums.com Webmaster (Site Usage Guide | Site Help | My Photography)



PentaxForums.com's high server and development costs are user-supported. You can help cover those costs by donating. Or, buy your photo gear from our affiliates, Adorama, B&H Photo, or Topaz Labs, and get FREE Marketplace access - click here to see how! Trusted Pentax retailers:

04-25-2014, 02:11 AM   #3
Veteran Member
zoolander's Avatar

Join Date: Dec 2012
Location: Brisbane
Photos: Albums
Posts: 334
Original Poster
QuoteOriginally posted by Adam Quote
You should easily be able to tell the difference between the two lenses- were you on a tripod when doing the tests? The whole AA filter thing was blown way out of proportion when filterless cameras came out and it really doesn't play a big role compared to resolution and dynamic range.
Yeah I was on a tripod, and the 18-135 at 50mm and 2 second timer. The 50 1.4 has less color etc
04-25-2014, 03:52 AM   #4
arv
Site Supporter




Join Date: Aug 2010
Photos: Albums
Posts: 374
I guess you should post 100% crops from centre and borders/corners. IMHO both lenses might be close in resolving power at f5.6 - f8.0, but the main advantage of fast lenses is ability to shoot at wider apertures.
Photozone did not test FA 50mm on K5, so these tests are not directly comparable.
Sharpening is a kind of bypass of AA filter - Photozone apply some sharpening (elaborated and equal for any specific camera, their use for testing), in their tests to show the best lens performance.
The main advantage of AA filter removal is just reduced demand for the sharpening.


A.

04-25-2014, 06:54 AM   #5
Pentaxian
Na Horuk's Avatar

Join Date: Feb 2012
Location: Slovenia, probably
Photos: Gallery | Albums
Posts: 10,325
Meh, the real problem is that "sharpness" as it gets measured doesn't really tell much about real world performance. In most situations you only need "enough" sharpness, beyond which differences are only theoretical.
The AA filter does make some difference, but its probably not earth-shattering. You can compare the K-5II and K-5IIs. There is a difference, if you use very good lenses and look very closely.
04-25-2014, 08:20 PM   #6
Veteran Member
zoolander's Avatar

Join Date: Dec 2012
Location: Brisbane
Photos: Albums
Posts: 334
Original Poster
QuoteOriginally posted by arv Quote
I guess you should post 100% crops from centre and borders/corners. IMHO both lenses might be close in resolving power at f5.6 - f8.0, but the main advantage of fast lenses is ability to shoot at wider apertures.
Photozone did not test FA 50mm on K5, so these tests are not directly comparable.
Sharpening is a kind of bypass of AA filter - Photozone apply some sharpening (elaborated and equal for any specific camera, their use for testing), in their tests to show the best lens performance.
The main advantage of AA filter removal is just reduced demand for the sharpening.


A.
I'll print out a test chart and perform some tests either today or tomorrow. But over the course of a year of owning the K-5, I have been wondering about this issue and haven't seen a significant difference in some of my lenses.

Adam posted a thread a while ago where he took a photo with a K-5iis and the K-3, he then asked the forum if we could pick the difference and which photo came from which camera. In that instance, the K-3's rendition of the water was nice and shiny, but K-5iis was dull and not shiny. I see this and believe this is due to the bayer pattern on the K-5 (unless a freak cloud caused shade !!). The K-5's seem to render out textures and 3 dimension-ality in a particular way - and this was one of the main reasons why I switched to Pentax .... its like a 3d machine ! But I'm wondering if that too could come into play.

The only time my 50's appear a little sharper is when I put on my UV/IR cut filter on them. I tested that filter out and found it did improved sharpness a little, but I got clobbered last time I made a tread discussing it.

But I'll do my tests on all my lenses and get back on this.

As for Photozone, their tests on the 10mp K-10 or 20 show distinct differences in resolution. The K-5 tests show an improvement. But when they reviewed the 43mm 1.9 they said it doesn't work on the K-5, which is funny because in the 43 reviews people use it on K-5's and don't have a problem. I think Photozone is just ramrodding some Pentax reviews because they hate Pentax, especially when it outperformed other brands at 10 megapixels (Hey but thats just my impression and opinion, and I've lost respect for a number of online reviewers).

I have no intention of modifying my AA filter.
04-25-2014, 11:03 PM   #7
Veteran Member
zoolander's Avatar

Join Date: Dec 2012
Location: Brisbane
Photos: Albums
Posts: 334
Original Poster
Finally:

Okay so I retested using a chart I found off of the net. I tested with the IR remote and mirror lock up, and got mixed results. This was because I didn't leave around 4 seconds to open the shutter. (I tested 18-135, FA 50 1.4 and F 35-70macro all @50mm)

I redid the tests with 2 second timer Mup, and Mirror lock up at least 4 seconds with IR remote. Again, 2 second timer was showing mixed results. But using Mup and a long 4 second delay yielded repeatable results.

So I learned something today, 2 second Mup is not that good and inconsistant, and Mirror lockup with the IR remote is the only way to go on a tripod.

Yes the 18-135 is least sharp, the F 35-70macro is second, and the FA50mm 1.4 is considerably sharper (and maybe a smidge sharper with the UV/IR cut filter on it)

But you guys already know that. Sorry for putting up with me !
04-29-2014, 10:21 AM   #8
Site Supporter




Join Date: Mar 2008
Location: Prince George, BC
Photos: Gallery
Posts: 2,255
Just a note that for astrophotography where the slightest vibration will kill a shot, using mirror lock-up and an IR remote release is mandatory. I count out a 5-second delay between MUP and shutter release to ensure all vibrations have damped down.

Jack

04-29-2014, 03:46 PM   #9
Veteran Member
zoolander's Avatar

Join Date: Dec 2012
Location: Brisbane
Photos: Albums
Posts: 334
Original Poster
QuoteOriginally posted by jbinpg Quote
Just a note that for astrophotography where the slightest vibration will kill a shot, using mirror lock-up and an IR remote release is mandatory. I count out a 5-second delay between MUP and shutter release to ensure all vibrations have damped down.

Jack
Its the new rule of thumb !
04-29-2014, 05:24 PM   #10
Veteran Member
Imageman's Avatar

Join Date: Dec 2013
Photos: Albums
Posts: 461
This really is a complex issue and you've opened a can of worms.


In film days emulsions could resolve very fine details up to the resolution of any lens. a good standard or kit lens could resolve 90 line pairs per millimetre on the emulsion. I know I tested this. The best lenses could resolve beyond 150 line pairs per millimetre.


This very fine resolution on the film was s much due to film being an analogue medium as it was the quality of the emulsion.


With digital sensors comes a whole new ball game. Gone is the expectation that you will see all of the resolution your lens is capable of, for the following reason.


The sensor photosites are all equal and a fixed size.


Lets examine a full frame sensor, lets say its 6 megapixel, yes they did exist. This would be 3000 x 2000 photosites spread across the frame.


But these photosites are not pixels, the bayer array demands that 9 photosites are used to generate 1 real pixel, therefore across the frame 3000 must be divided by 3 to reveal the true resolution of this sensor.


3000/3=1000 therefore there are 1000 pixels per 35mm or 1000/35=29 pixels per mm.


The truth is that due to sophisticated computer processing in camera, detail is interpolated and this 29 pixels can effectively be doubled to around 60 pixels per mm.


Unfortunately to represent a line pair takes 2 pixels, so the true resolution of this sensor is back down to 30 line pairs per mm.


Unfortunately due to the need to reduce aliasing, the dreaded anti aliasing filter was introduced and further reduced this resolution down to, well anyones guess. 25 or so line pairs would be a good guess. It all depends on how strong the filter is.


Now lets consider a 24 megapixel full frame sensor


This sensor would be 6000 x 4000 photosites and the math is 6000/3=2000 2000/35=57 again this can be doubled for interpolation and then halved for line pairs, yielding 60 line pair resolution, and with the lack of an anti aliasing filter we would have true resolution of the sensor able to discriminate between lenses of 45 to 60 line pairs. But if you put a lens on that sensor with any more resolution than 60 line pairs, you will see no difference in resolution whatsoever.


Now finally lets discuss the k5 with 23.4mm x 15.6mm sensor size and 4928 x 3264 photosites


The math is 4928/3=1642 1642/23.4=70, this effectively represents 70 line pairs per mm and the anti aliasing filter will reduce this to possibly 60 line pairs per mm.


This is enough resolution to distinguish between good lenses but is not in my opinion outstanding. It is however a solid performance.


So where does all this leave us?


Well all this is really saying is yes there is a limiting factor governing lens resolution but the anti aliasing filter is not it.


The photosite size and pixel size is the true limiting factor to lens performance and you should consider this when matching lenses to digital cameras.


Do this math on the sensor specs sensor horizontal photosites / 3 / sensor width in millimetres = line pairs the sensor can resolve


This should equate to Nyquist, but unfortunately due to variances in interpolation in camera and the AA filter, there will be observable differences.


I hope this hasnt confused too much but it is as I said at the start a complex issue.


Interestingly and slightly off topic, these figures suggest that if we move to a more expensive full frame camera from APS-c with exactly the same megapixels, the photosite size must be significantly larger and therefore cheaper lenses can be used with little perceivable impact on true resolution, Opening the door maybe to controlling or reducing significantly the costs of the change to full frame.
04-29-2014, 06:57 PM   #11
Veteran Member
zoolander's Avatar

Join Date: Dec 2012
Location: Brisbane
Photos: Albums
Posts: 334
Original Poster
QuoteOriginally posted by Imageman Quote
Interestingly and slightly off topic, these figures suggest that if we move to a more expensive full frame camera from APS-c with exactly the same megapixels, the photosite size must be significantly larger and therefore cheaper lenses can be used with little perceivable impact on true resolution, Opening the door maybe to controlling or reducing significantly the costs of the change to full frame.
I found your whole response interesting. It kind of sounds a little "theoretical", because the removal of the AA filter in the K-5iis has made an improvement. Plus the K-3 seems to out sharpen the Nikon D600, and I've even seen a blog where the K-3 appears to utterly out sharpen a D800.

LK1:1photoGRAPHICS: Nikon D800 and Pentax K3 / Zeiss 50mm f1.4 and Pentax 43mm f1.9

That being said, I'd attribute this to the AA filter being fairly strong on the D800 as compared to the K-3. Plus the 43mm 1.9 is better.

All in all the math and the theory all adds up. But every manufacturer uses different AA filters.

So theoretically, the future Pentax full frame might show what you just discussed. But I don't think it may be applied across manufacturers, perhaps with the top tier. But I'm no expert, and I see no real reason to move to full frame when my K-5 has excellent IQ comparable to some top tier full frames (except high ISO's). There's no need to throw a truck load of money on a Canon or Nikon FF just for ISO 6400. I can spend $500 on a Fujifilm X-A1 if I want to print at ISO 6400 and 12800. This camera appears to dispel some of the APSC low light myths.

I guess we'll have to see how the Pentax full frame camera performs.

I'm thinking that the full frame manufacturers will probably pull back and put on fairly strong AA filters, to dumb down (blur up) FF kit lenses. If the AA filter is good, and conversely the FF kit lenses are nice and sharp, then there's very little incentive for a consumer to pay a truck load of money for better glass. They'd kill the business model. I think this is apparent with the D600 and D800, but I'm not sure how sharp the 6D or 5Diii.

So I'm thinking that entry level Full Frame, is just like entry level APS-C camera. And Top tier APS-C is better than entry level Full Frame. Its a very greedy business model that contradicts the theory and math.

Last edited by zoolander; 04-29-2014 at 07:10 PM. Reason: Adding more thoughts
04-30-2014, 06:38 AM   #12
Veteran Member
Imageman's Avatar

Join Date: Dec 2013
Photos: Albums
Posts: 461
I have to respond,


I think you should read my response again I think you may have misunderstood slightly.


You suggest it sounds theoretical, im sorry you feel this, maybe I didn't explain well, it is after all a highly complex topic.


A theory is an unproven speculation, so for my post to be theoretical it must be speculative, None of my post apart from some anti alias filter assumptions were speculation.


Sensors are a collection of light sensitive photosites, these photosites are all the same size, they are grouped together in pixels, the pixel sizes can be worked out, the pixel size limits the resolution of any lens attached. None of this is speculative. This is all fact and cannot be disputed.


I also explained how to work out the physical size of the pixels each sensor uses to capture details the lens resolves based on the sensor specification.


This allows you to directly compare the real resolving power of any sensor, and I gave 3 examples one of which was for the k5.


This way of comparing sensor resolution is the only way to do so in absolute terms, the other way to compare is to look at the output and that's unscientific and open to interpretation.


Have another read of it im sure it will become clearer and seem more factually accurate when you have.


I challenge you to use my method to compare the D800 and k3 rather than trust the comparison you have linked which I found to be lacking in any meaningful comparison of absolute sharpness.


I need to clarify something. In my original post I made no reference to image sharpness across the frame from one side to the other, this is the usual measure of overall sharpness. I limited myself to pixel sharpness because this is the factor that determines how lens resolution is recorded in the image.


Its confusing as there are 2 different sharpnesses at work here pixel sharpness and image sharpness so I need to explain.


Lets assume we have a sensor that is 1/4 of the area of a full frame sensor, that means it is half the width and half the depth. Lets further assume that the smaller sensor has 15 megapixels and the full frame sensor 20 megapixels.


On paper the smaller sensor is the sharpest, the pixel size is significantly smaller than on the full frame and it should generate the sharpest image when examining pixels alone, they are after all smaller and tighter packed on the sensor and more finer detail can be recorded, however the full frame sensor has more megapixels in the image and in fact generates the sharper image of the two when examined as an image or printed up.


In pixel sharpness and recording lens resolution smaller pixels win, in image sharpness more pixels win.


Of course a full frame sensor with a huge number of tightly packed photosites is the winner in all areas but that would require 120 megapixels, and until they arrive, everything is a compromise.


In your reply to my post you talk of comparing the sharpness of the k3 to the D800, this appears to be an image sharpness comparison your making, unrelated to sensor/pixel sharpness and that's perfectly valid and appropriate. I simply suggest that you consider whether your comparing pixel sharpness or image sharpness and the comparison will I think be more meaningful.

At the end of the day, my post only addressed your topic which was, will your sensor reveal sharpness differences between two lenses, and I answered that in explaining how individual pixel sizes vary on sensors and how they relate to lens resolution.


Had your post asked about image resolution we would have been forced along an even more complex path that I have merely touched upon so far.


And we haven't even considered lens performance across the frame and how variations in resolution across the frame are affected by a crop sensor compared to a full frame.


Nor have we considered Bayer array sensor resolutions compared to Fovion sensor resolution


Nor have we considered sharpening in post processing which affects perceived image sharpness and can be applied more aggressively on some sensors


Another huge post, but I said this is a can of worms, and its a huge can.
05-01-2014, 04:41 AM   #13
Veteran Member
zoolander's Avatar

Join Date: Dec 2012
Location: Brisbane
Photos: Albums
Posts: 334
Original Poster
You wrote this:

QuoteOriginally posted by Imageman Quote
The math is 4928/3=1642 1642/23.4=70, this effectively represents 70 line pairs per mm and the anti aliasing filter will reduce this to possibly 60 line pairs per mm.


This is enough resolution to distinguish between good lenses but is not in my opinion outstanding. It is however a solid performance.


So where does all this leave us?


Well all this is really saying is yes there is a limiting factor governing lens resolution but the anti aliasing filter is not it.
You cannot discount the limiting factor of the anti aliasing filter based on speculation.

A k-3 with no anti-aliasing filter versus a D600 with anti-aliasing filter (of indeterminable blur) appears to be clearly sharper.

Nikon D600 vs. Pentax K-3 Image Comparison: Pentax Sweeps Nikon

Its the AA filter. Plus with the post for the K-3 vs the D800, again its the AA filter and the K-3's superior image processing engine (which you've also said contributes to sharpness).

At the end of the day, a picture paints a thousand words - and thats why I'm with Pentax, and I wouldn't throw away a truck load of money on a D600 or D800.
05-01-2014, 07:51 AM   #14
Veteran Member
Imageman's Avatar

Join Date: Dec 2013
Photos: Albums
Posts: 461
Why do you insist on misunderstanding and misquoting me.


I never suggested that the antialiasing filter does not affect sharpness I said that it does. I believe I proved that pixel sharpness is the limiting factor between lenses and their resolution.


You started this thread off asking about one kind of sharpness the kind that can reveal the difference between lenses, this is pixel sharpness. I answered that, and now you shift your ground and start talking about image sharpness, quite a different thing and my second post touched on that.


You have now included a link to yet another comparison between cameras this time a D600 and a k3, and its another unscientific comparison.


The Nikon sported a nikkor lens, the Pentax a pentax lens, not even the same lens, how is this a comparison. Its as much a comparison between lenses as between cameras. Is it really a surprise that the results differ. How can the difference be attributed as you do to the AA filter or lack of it.


The Lenses are different, the sensor photosite sizes are different, the AA filter is different, the processing engine is different.


What I will say is, in the context of what you now see as important the k3 image is the sharper, and this is partly due to the lack of an antialiasing filter.


But you are seemingly fixated on the antialiasing filter or lack of it as the only real factor affecting sharpness, and your just plain wrong.


Have you tried as I suggested, comparing the sensors using the method I proposed, all you seem to have done is trawled the net looking for proof of your argument.


The comparisons between cameras that you come up with and you use to suggest that the antialiasing filter is the difference in sharpness between them, is not an apples oranges comparison, its an apples motorbikes comparison.


Factors affecting sharpness in an image are:-


1 photosite size
2 pixel size
3 anti aliasing filter
4 processing engine
5 lens resolution
6 contrast
7 in camera sharpening
8 camera shake
9 post processing sharpening


Don't get me wrong here I completely agree that removing the anti aliasing filter does yield an increase in resolution and that's why for the last 8 years I have argued for the abandonment of antialiasing filters. I went so far as to create 7 years ago a photoshop workflow to remove aliasing quickly and easily if it ever appears. It never has.


But really there are many more factors affecting sharpness than this antialiasing filter.


If you want a scientific test that shows the real difference between images with and without anti aliasing filter, concentrate on those cameras with that option of AA filter removal. the D800 and d800e for example or the k5 II and k5 IIs. look at the difference between images where the only difference is the filter removal. Same subject same camera same lens same sensor.


I have done this comparison and there is a difference but its not as profound as you suggest.
05-01-2014, 08:21 AM   #15
Pentaxian
kh1234567890's Avatar

Join Date: Sep 2010
Location: Manchester, UK
Photos: Gallery
Posts: 2,327
QuoteOriginally posted by Imageman Quote
If you want a scientific test that shows the real difference between images with and without anti aliasing filter, concentrate on those cameras with that option of AA filter removal. the D800 and d800e for example or the k5 II and k5 IIs. look at the difference between images where the only difference is the filter removal. Same subject same camera same lens same sensor.

I have done this comparison and there is a difference but its not as profound as you suggest.
I could never find any real life comparison shots between the K-5ii and K-5iis - beside the contrived testing site ones, carefully lit and taken with a lens sharper than anything that I have. And on those the difference is absolutely minimal.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
anti-aliasing, fa, filter, k-5, k-5 anti-aliasing filter, lenses, lot, pentax help, sharpness
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
II or IIs -- Anti-aliasing question jon404 Pentax K-5 9 03-05-2014 11:05 AM
Anti Aliasing help RockvilleBob Pentax DSLR Discussion 1 10-01-2012 10:38 AM
Can the anti aliasing filter be removed from the K5 MikeD Pentax K-5 10 09-11-2012 10:23 PM
anti-Aliasing filter for 7D Anvh Non-Pentax Cameras: Canon, Nikon, etc. 3 06-30-2012 10:17 AM
Pentax please REMOVE Anti-Aliasing on K-7 LCD illdefined Pentax DSLR Discussion 22 09-20-2009 02:44 PM



All times are GMT -7. The time now is 06:04 PM. | See also: NikonForums.com, part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top