Home · Register · Search · View Winners · Software · Hosting · Software · Join Upload & Sell

Moderated by: Fred Miranda
Username   Password

  New fredmiranda.com Mobile Site
  New Feature: SMS Notification alert
  New Feature: Buy & Sell Watchlist
  

FM Forums | Alternative Gear & Lenses | Join Upload & Sell

1       2       3              25      
26
       27              43       44       end
  

Archive 2011 · Still no love for the Ricoh GXR?
  
 
theSuede
Offline
• • • •
Upload & Sell: Off
p.26 #1 · Still no love for the Ricoh GXR?


Kosmos; Well, then lets hope that the M-module sells out quickly then... :-)

Yes, the difference between 12 and 16MP is actually quite a lot - in other areas it might not amount to much, but regarding pattern sensitivity it is. Even better would be if Ricoh can opt for the 24MP version now (apparently) arriving in the Nikon D3200 too.



Feb 06, 2012 at 10:46 PM
atran
Offline
• •
Upload & Sell: Off
p.26 #2 · Still no love for the Ricoh GXR?


theSuede wrote:
Kosmos; Well, then lets hope that the M-module sells out quickly then... :-)

Yes, the difference between 12 and 16MP is actually quite a lot - in other areas it might not amount to much, but regarding pattern sensitivity it is. Even better would be if Ricoh can opt for the 24MP version now (apparently) arriving in the Nikon D3200 too.

I would say the stock is running out in the US the M-module is is back-ordered at both BH and Adorama. There is some stocks at PopFlash though



Feb 06, 2012 at 11:28 PM
Kit Laughlin
Online
• • • •
Upload & Sell: On
p.26 #3 · Still no love for the Ricoh GXR?


theSuede wrote:

As long as you have no moire, and don't care about reality, this is good on a print (most of the time).

So, what kind of unreal artifacts are you talking about, and if they can't be seen (mostly) at a print level, I wonder how important they are?

So, two questions: please elaborate on the kind of unrealities that are introduced by a non-AA camera (just for the sake of understanding).

And the second question is, could you discuss what you mean by balancing sharpening (in my workflow, if I do sharpen, it is always the last thing I do, and often just a touch of local sharpening on a layer via a brush; this is not real either, of course, but it can help an image achieve its purpose). I would be very grateful for your suggestions about this, as well as any comment re. convolution sharpening vs. any others you recommend. TIA, KL



Feb 07, 2012 at 01:31 AM
kosmoskatten
Offline
• • • •
Upload & Sell: Off
p.26 #4 · Still no love for the Ricoh GXR?


theSuede wrote:
In a camera without an AA filter some of all this is already taken care of before the interpolation stage, before all of the curve-modifying stages, which makes the image "easier to handle" regarding sharpening in post.

and...

Sharpening an AA image after development does not sharpen structures and bright areas nearly as much as it "should" - or more accurately, it oversharpens darker details so much that you have to keep the sharpening very low in strength to not make some areas of the image look viciously oversharpened. There are some very easy ways around this, when you learn
...Show more

This is my main point. Also I find that the AA filter implementation in many cameras is not as good as it could be making them unnecessarily prone to the waxy look that I don't like and harder to work with in post.



Feb 07, 2012 at 05:29 AM
theSuede
Offline
• • • •
Upload & Sell: Off
p.26 #5 · Still no love for the Ricoh GXR?


Kit Laughlin wrote:
So, what kind of unreal artifacts are you talking about, and if they can't be seen (mostly) at a print level, I wonder how important they are?

So, two questions: please elaborate on the kind of unrealities that are introduced by a non-AA camera (just for the sake of understanding).

And the second question is, could you discuss what you mean by balancing sharpening (in my workflow, if I do sharpen, it is always the last thing I do, and often just a touch of local sharpening on a layer via a brush; this is not real either, of course, but
...Show more

1) mostly dithering effects, and "pixel placement" deficiencies. You see it as "grit" in a normal image, almost any image. I see it as crap, some people see it as "detail". I have a hard time accepting as "detail" since it isn't there in the reality in front of the lens. It mostly disappears when you use the image at downsampled resolutions, sizes smaller than ~40%.

2) I'd be happy to. But that's a topic for the "post processing & printing" forum part. If I have the time to put together a tutorial later tonight, I will do that and add a link here.



Feb 07, 2012 at 11:12 AM
Kit Laughlin
Online
• • • •
Upload & Sell: On
p.26 #6 · Still no love for the Ricoh GXR?


to theSuede:

re. 1) If this is random enough (relative to the frequencies of the parts of the image in which it occurs), then is there really a problem? What I am thinking about here is, as these artifacts do not appear/nor are visible as something different to the reality being depicted, especially when downsampled, then what is the concern? In the process of converting numbers to pixels in the raw processing process, we do not see individual pixels, let alone their individual placements—only their aggregated effects—so if these artifacts are sufficiently random, then likely not a 'problem'. Moiré, as you mentioned before as well, is an example where, at viewing/printing scale the artifacts *are* a problem, just because their effects are not seen as random. Apart from moiré, are any of these artifacts visible at printing scales?

re. 2) I (and I am sure many others here) will appreciate that, sincerely. We are all here to learn.



Feb 07, 2012 at 11:58 AM
theSuede
Offline
• • • •
Upload & Sell: Off
p.26 #7 · Still no love for the Ricoh GXR?


No, the thing is that they're not random in nature. The "grit" appears on smaller edges and lines, and color interpolation faults appear on every line that's not wider than 2 pixel-widths. When you blur chroma ("fix moire and aliasing"), those color faults do not translate into the right brightness value, and you get "dot-slash" lines. Lines that should be continuous, but are broken into alternating segments of too bright - too dark.

In a print that doesn't have to up-sample the given resolution this is no obvious problem. The grain of the ink-dots will dither the faults into invisibility most of the time, if they aren't to obvious.

Those faults (but not moire) also often average out really nicely when you downsample the image - but if you downsample, why the AA-less image in the first place?

Being without an AA filter can be motivated by angle-response needs, like the ones you have in a symmetrical wide angle lens. But as I see it it can never be a substitute for having actual, real target resolution there in the first place. If you NEED a perfect 10MP image, your original should be at least 20MP. Basically, these are the same rules of thumb that once went into the old analog screening process in the print-process.



Feb 08, 2012 at 01:02 AM
Kit Laughlin
Online
• • • •
Upload & Sell: On
p.26 #8 · Still no love for the Ricoh GXR?


That was very helpful, thank you. And if you ever post on the sharpening process you mention above, please post a link to it here or PM me. Cheers, KL


Feb 08, 2012 at 07:14 AM
kosmoskatten
Offline
• • • •
Upload & Sell: Off
p.26 #9 · Still no love for the Ricoh GXR?


Thanks Joakim, it is interesting to get the technical explanation behind some of the phenomena I have seen while printing digital over the years and that the explanation is somewhere along the line as to what I thought had caused it.


Feb 08, 2012 at 06:07 PM
denoir
Offline
• • • •
Upload & Sell: Off
p.26 #10 · Still no love for the Ricoh GXR?


Kit Laughlin wrote:
So, two questions: please elaborate on the kind of unrealities that are introduced by a non-AA camera (just for the sake of understanding).


It has to do with when you are sampling a periodic pattern. If the sampling frequency is too low (i.e pixel density in this case) you'll just get part of the pattern and it won't be an accurate representation of reality. There are two different schools of thought when it comes to this, Joakim (theSuede) belongs to one and I to the other. People in his camp prefer not to show anything than to produce an inaccurate representation while people in my camp say that partial information is better than none, even if inaccurate.

I can give you a very simple practical example. Here is a closeup of a window taken with a long tele lens. Note the frequency of the slats in the window blinds, and also note the ventilation shaft on the left:







Here are two images (100% crops) both taken with a Leica M9 (no-AA filter). Shot B however uses a very small aperture so we get through diffraction the same effect an AA filter would produce:







Now, in shot A, look at the blinds. Can you see that something doesn't add up? The slats in the blinds are far too few. And the same goes for the horizontal openings in the ventilation shaft. This is what you get when you get rid of the AA filter.

The two possible ways of looking at it is:

1) Those are false images and don't represent reality. It's better not to show anything than to fabricate.

2) In image A you can actually see that there are blinds in the window and that there's a ventilation shaft. Even if it isn't represented 100% accurately it still conveys much more information than if you just remove those elements.



Feb 08, 2012 at 06:56 PM
 

Search in Used Dept. 



kosmoskatten
Offline
• • • •
Upload & Sell: Off
p.26 #11 · Still no love for the Ricoh GXR?


Denoir: then I am definitely in your camp. That was an excellent example, thank you.

I think the technical delving into how good a technical solution is or is not - on paper - not always rhymes with how it works in reality, or should we say _practical_ reality.

With the ink diffusion/dithering and the final printed image in mind I find that the GXR M does produce a very well perceptively sharp and detailed image with minimal fuss and post pro without introducing perceptively distracting artifacts. On the few occasions where artifacts have snuck in they have either been easily addressed or not showing in final print and I have yet to find a situation where the lack of AA-filter has been a real obstacle.
That's what counts in my book.

(Don't ask me to photograph a group shot of men wearing kilts... )




Feb 08, 2012 at 07:29 PM
michaelwatkins
Offline
• • •
Upload & Sell: Off
p.26 #12 · Still no love for the Ricoh GXR?


I am in the camp of practical reality too. Having used two comparable cameras albeit one with a higher resolution sensor, with the same suite of lenses, I would today again take the lower resolution AA-filter-less camera again. Maybe in other comparisons I'd not come to that preference - my supposition is that the actual implementation of the AA filter makes a great deal of difference.

Or maybe the preference is based on something baser... perhaps I prefer grit over plastic?



Feb 08, 2012 at 09:39 PM
bluetsunami
Offline
• • •
Upload & Sell: Off
p.26 #13 · Still no love for the Ricoh GXR?


Unless an AA less body has the ability to capture two nearly identical images of my niece in the same frame, one from a parallel universe I'll be fine with false detail at the smallest of levels


Feb 08, 2012 at 09:54 PM
Mitch Alland
Offline
• •
Upload & Sell: Off
p.26 #14 · Still no love for the Ricoh GXR?


I have been following the above discussion on AA filters with as much interest as would a medieval plowman a canonical disputation on how many angels can stand on the head of a pin. Also, a joke comes to mind about a man who comes home and finds his wife naked in bed with another man: she says, "Darling, what are you going to believe what I tell you or your eyes?"

I believe my eyes. The bottom line for me is that files from the M8.2, M9 and GXR-M look substantially better and print better than files from cameras with AA filters, and (while I'm at it) that the the GXR-M has better color accuracy than the M8.2 and the M9, and, yes, that developing M8.2, M9 and GXR-M files with RPP results in better resolution and color rendition than with other raw developers that I've used (Aperture, Lightroom, Capture One, Raw Developer, Lightzone and Silkypix).

—Mitch/Pak Nam Pran
Pak Nam Pran



Feb 08, 2012 at 10:46 PM
theSuede
Offline
• • • •
Upload & Sell: Off
p.26 #15 · Still no love for the Ricoh GXR?


Denoir;
1) what you MIGHT want to try is to shoot the same lens on the 5N and the M9 at the same aperture and shutter speed... Sharpen the 5N image, and the downsample it to M9 object scale. OK, you won't get the false detail you like, but check out the detail increase and lack of edge aliasing... The 5N kills the M9 scale for scale. Half the noise, more resolution, 10x less artifacting. But unfortunately on an APS sensor scale...
(you might want to take a reference shot with either the 5N or the M9 with a lens at least twice the FL, so that you get a higher magnification "reality check" on the target. So that you know what should be there, and what shouldn't)

2) I assume you stopped down, and used a longer shutter speed in image section "B". Here you can see another very distinct reason to use an AA filter for almost all normal camera users; The "noise" increases by a factor of 2 when you remove the AA filter.

The "noise" in the image without filter is not really "noise" in the way most people think about it - it's a perfect measure of the amount of false detail, dither, aliasing and color fault you get added to EVERY edge contrast in the picture. It shows fairly exactly the amount of "interpolation mistakes" the raw-converter does when confronted with an impossible task, to interpolate where there's no information (sparse sampling).
The noise increase shows the minimum amount of false detail present on EVERY contrast edge in the image, no exception.

We agree to disagree on the "use" of this image defect... :-)



Feb 08, 2012 at 11:28 PM
denoir
Offline
• • • •
Upload & Sell: Off
p.26 #16 · Still no love for the Ricoh GXR?


theSuede, I've done such comparisons, mostly with the C3 which has the same sensor as the 5N.

Here's one example - with a 100% M9 crop and a reduced C3 crop matched to the same FOV. The lens used in the test was the Zeiss 25/2.8 Biogon ZM. It holds the distinction of being the lens with highest recorded resolving power (400 lp/mm, MTF10):







What is actual detail and what is noise or false detail? Difficult to tell. The AA filter on the NEX reduces the micro contrast significantly so while the detail may be it might blend together too much with the background to be noticeable. Of course the AA filter is not the only culprit - the higher pixel density of the NEX means lower MTF for the same detail.



Feb 09, 2012 at 12:07 AM
denoir
Offline
• • • •
Upload & Sell: Off
p.26 #17 · Still no love for the Ricoh GXR?



Of course, ideally you would have a gazillion LARGE pixels on a sensor which you could then downsample to a reasonable size. Then AA or no AA wouldn't matter nor would really the lens (as far as raw optical performance goes).

Here's an old example - a 396 image (7 Gigapixel) @ ISO800 stitched pano with one of the worst cheap super zoom lenses ever made vs a single wide angle shot from the Zeiss 25/2.8 Biogon @ ISO160.


It is impossible to describe in words how bad the Sigma lens is. It's an old consumer grade super zoom - it really can't get worse. I keep it around because I can on occasion use it for test shots like this one and the alternative would have been to chuck it into the nearest garbage bin. While it's best performance is at 200mm, which is what I shot at, you can see from the following 100% crop just how bad it is:








Anyway, Wide angle shot:







396 image Pano:






And now to the results and the goal of this test - how do the images compare when the pano is resized to the size of the single WA shot:

100% crop near the center of the image:
WA on the left, pano on the right:







(Left WA, right Pano). It's at the extreme right edge of the image:






Of course, you don't have to go to such extremes. If you use a half-decent lens you can get away with just a couple of shots:
Here's a quick and dirty test of getting a good zoom lens (5D2) match the resolution of an excellent prime (M9). I shot this handheld and the stitching isn't the best, but it doesn't matter for the test as we'll be looking at center resolution.

The zoom in question is the Canon 16-35/II used with a 5DII and the prime in question is the Zeiss 18/4 ZM used with an M9.

Single shot ZM18:






Single shot 16-35:






4 image Pano:







100% Crop comparison when the pano is resized to the native size of the M9:








Feb 09, 2012 at 12:21 AM
theSuede
Offline
• • • •
Upload & Sell: Off
p.26 #18 · Still no love for the Ricoh GXR?


Thought you had the 5N? No matter. I find that it is quite up to the person behind the wacom to decide how such a comparison should end...

I feel deeply sorry for the poor unfortunate souls that live in the building to the left, the contractor obviously used scrap material for the entire project. None of the window-sills or frames have the same measurements / thicknesses. Doesn't bode well for the internal structures...

Starting from your web-compressed jpg, spending five (well let's not lie now - I spent at lest seven-eight, but anyway) minutes on finding the right curve reversals for the C3 part I came up with this. The result - if I started from raw - would take considerably less time, and also end up including considerably more detail and sharpness (and of course also less sharpened jpg-artefacts).





Feb 09, 2012 at 01:07 AM
theSuede
Offline
• • • •
Upload & Sell: Off
p.26 #19 · Still no love for the Ricoh GXR?


With your image, as you presented it as reference:




Feb 09, 2012 at 01:10 AM
theSuede
Offline
• • • •
Upload & Sell: Off
p.26 #20 · Still no love for the Ricoh GXR?


Some color profiles obviously didn't translate right, sorry about the contrast / saturation increase in my resave of the C3 version - I meant to keep your reference M9 image constant.

[edit]
We are now comparing a 400 (four hundred) € camera to a 6000 (six thousand) € camera.



Feb 09, 2012 at 01:13 AM
1       2       3              25      
26
       27              43       44       end




FM Forums | Alternative Gear & Lenses | Join Upload & Sell

1       2       3              25      
26
       27              43       44       end
    
 

You are not logged in. Login or Register

Username   Password    Retrive password