Moirée DP2M

Try using nohalo or lohalo instead of cubic, when you scale down the image from the original size.
 
How does it look taken from a high resolution tiff to a print? Any digital device will produce Moire because they are digital devices and are governed by the fact without anti-aliasing filter will show aliasing. A digital monitor is a digital device. My guess is some of this is related to the fact it is sized to be displayed on a digital device on the Internet.

However, such patterned fabric is very prone to Moire because of the high spatial frequency content. We now the Foveon will display Moire since it has no anti-alising filter. What would be really interesting would be to see this same image using the Q sensor.
 
Why would you want to remove that? It looks SO cool.

;)

I thought Merrill cameras were incapable of producing moiré (unless it is something happening in real life). I guess not, huh?
I think it's colour moire that the Merrills (and other true Foveons) avoid. Any sensor with no AA filter and paired with an ultrasharp lens can suffer from aliasing, which is what causes moire patterns. Mind you, I think other forces are at play here!
 
That has to be a re-sizing artefact or can you show us the full size original?
If it turns out to be a re-sizing artifact, which does seem likely, try applying a small amount of Gaussian blur before down-sampling, maybe 0.7px? The greater percentage of down-sampling, the more the blur amount needed.

How about posting a crop from a 100% JPEG? Then we can all play with it :-)
Not forthcoming, apparently?
 
If resized, the resizing is likely the reason.

If crop, then it certainly looks ugly. Softer lens, tiny aperture (due to diffraction), vaseline on the lens, different distance to the offending subject all can help. Or using a camera with an anti-alias filter. Since there are AA, hotmirror etc. removal services, I wonder if there is AA-adding service (may change the optical stack height though so possibly PDAF and image edge quality issues).
 
mmm... I think you use full res TiFF or DNG for your test... So when you are in front of your screen and you zoom at 100% you see that kind of moiré?

It's new for me because shooting fabrics with DP2 and 3 I never have this problem. I also shoot with "no AA filter" cam such as Ricoh GR and PEntax K3 and that's the kind of problem I sometimes encounter.

Anyway you're more aware of this kind of things because you test a lot of different cam...

But for the submited shoot at the first post I'd be curious to have the original file. Because the really few times I had this behaviour on fabrics with DP, it was because of a "bad resizing". I really never had this on a DNG in the same situation... curious... :-/
 
That's not moire, that's how the shirt was designed, look at the collar.

--
“The most puzzling development in politics during the last decade is the apparent determination of Western European leaders to re-create the Soviet Union in Western Europe.”
Mikhail Gorbachev
Tony
http://the-random-photographer.blogspot.com/
If that's the way the shirt is supposed to look . . . it is one CRAZY shirt! lol
That's what I thought, too; I'd buy one! :D Moiré shirts could be a new trend ... maybe one should offer color moiré as well. 8-)
 
I thought Merrill cameras were incapable of producing moiré (unless it is something happening in real life). I guess not, huh?
I think it's colour moire that the Merrills (and other true Foveons) avoid. Any sensor with no AA filter and paired with an ultrasharp lens can suffer from aliasing, which is what causes moire patterns. Mind you, I think other forces are at play here!

And I think he's perfectly right. Here's a real world example from DP3M with a grid/lattice/trellis: http://wxyz.de/bilder/orte/Goettingen/DP3M0083_o.jpg See the right tower at its top left. U can see the grid from straight, but a little more left and tilted, the meshes become narrower ...
 
How does it look taken from a high resolution tiff to a print? Any digital device will produce Moire because they are digital devices and are governed by the fact without anti-aliasing filter will show aliasing. A digital monitor is a digital device. My guess is some of this is related to the fact it is sized to be displayed on a digital device on the Internet.

However, such patterned fabric is very prone to Moire because of the high spatial frequency content. We now the Foveon will display Moire since it has no anti-alising filter. What would be really interesting would be to see this same image using the Q sensor.
 
I thought Merrill cameras were incapable of producing moiré (unless it is something happening in real life). I guess not, huh?
I think it's colour moire that the Merrills (and other true Foveons) avoid. Any sensor with no AA filter and paired with an ultrasharp lens can suffer from aliasing, which is what causes moire patterns. Mind you, I think other forces are at play here!
http://www.dpreview.com/forums/post/54263067

And I think he's perfectly right. Here's a real world example from DP3M with a grid/lattice/trellis: http://wxyz.de/bilder/orte/Goettingen/DP3M0083_o.jpg See the right tower at its top left. U can see the grid from straight, but a little more left and tilted, the meshes become narrower ...
Well, you can see patterns with in the cables of a large bridge as well, but then you also see the asame with your eyes. See this photo



Rio-Antirio_bridge_cropped.jpg






It is not taken with a DP, but I have taken photos of the same bridge with both DP2 and DP2M and you cannot avoid the effect, but then you also see the exact same effect with your bare eyes.



--
Out lightwriting with Sigma dp2Q
August feast 2014 - one photo each day
 
I thought Merrill cameras were incapable of producing moiré (unless it is something happening in real life). I guess not, huh?
I think it's colour moire that the Merrills (and other true Foveons) avoid. Any sensor with no AA filter and paired with an ultrasharp lens can suffer from aliasing, which is what causes moire patterns. Mind you, I think other forces are at play here!
http://www.dpreview.com/forums/post/54263067

And I think he's perfectly right. Here's a real world example from DP3M with a grid/lattice/trellis: http://wxyz.de/bilder/orte/Goettingen/DP3M0083_o.jpg See the right tower at its top left. U can see the grid from straight, but a little more left and tilted, the meshes become narrower ...
Thanks, Ritter, excellent example of Foveon moire!

Here's the Foveon paper on the subject:



--
Cheers,
Ted
 
mmm... I think you use full res TiFF or DNG for your test... So when you are in front of your screen and you zoom at 100% you see that kind of moiré?
TIFF.

If you go back and look again you will see that the images are zoomed 400% in FastStone Viewer. This shows certainly that the image moiré is caused by the sensor's geometric resolution and is not the moiré you can get on the screen when zooming in/out.

Moiré from repeating patterns in a scene become possible above a certain frequency of repetition, called Nyquist. In the simple case, think of a lens at infinite focus. The focus distance to the sensor for a DP2M is 30mm. One sensor pixel is 5um from the next one (horizontally) up to 7.07um (diagonally).

So any repeating pattern in the scene must not be closer together (at the sensor) than those numbers. If we shoot a shirt at 3m distance (3000mm) that is a factor of 100 times the focus distance. So . . . the threads must not be closer than 5um x 100 = 0.5mm (horizontally).

--
Cheers,
Ted
 
Last edited:
In reply to all the posters taking time to look at the problem I'm supplying these pictures that support the idea that the problem occurs when downsizing i.e. picture not rendered at 100%. The artefacts are definitely not a (psychedelic) pattern on the shirt.



Image at 100%
Image at 100%



Image at 82%
Image at 82%



Image at 23%
Image at 23%
 
In reply to all the posters taking time to look at the problem I'm supplying these pictures that support the idea that the problem occurs when downsizing i.e. picture not rendered at 100%. The artefacts are definitely not a (psychedelic) pattern on the shirt.
It's what most of us thought and guessed. The recommendation with blurring first is one possibility. The other one is using another algoritm for resampling. Below is cubic-spline ... 82% + 32% ... No moirè, but quite soft. It's always a tradeoff between sharpness and moiré ...

Cubic-Spline with IM (-filter Cubic -define filter:c=.25 -distort resize x%)
Cubic-Spline with IM (-filter Cubic -define filter:c=.25 -distort resize x%)
 
Last edited:
I thought Merrill cameras were incapable of producing moiré (unless it is something happening in real life). I guess not, huh?
I think it's colour moire that the Merrills (and other true Foveons) avoid. Any sensor with no AA filter and paired with an ultrasharp lens can suffer from aliasing, which is what causes moire patterns. Mind you, I think other forces are at play here!
http://www.dpreview.com/forums/post/54263067

And I think he's perfectly right. Here's a real world example from DP3M with a grid/lattice/trellis: http://wxyz.de/bilder/orte/Goettingen/DP3M0083_o.jpg See the right tower at its top left. U can see the grid from straight, but a little more left and tilted, the meshes become narrower ...
Damn that's annoying! (I mean those moiré patterns.) I'm SURE those were not visible by the naked eye. So sad. We need a higher resolution Foveon sensor, obviously! ;)
 
The following are resized using different scaling methods from that larger 100% size crop of the patterned shirt:



Cubic downscaling sample.
Cubic downscaling sample.



Lohalo downscaling sample.
Lohalo downscaling sample.



Nohalo downscaling sample.
Nohalo downscaling sample.

I used GIMP 2.9 to create these from this sample image:



The image from above, in this thread, which I used to create my downscaled samples.
The image from above, in this thread, which I used to create my downscaled samples.

As you can tell, a different method of downscaling can make all the difference. The same goes for upscaling. Some methods of upscaling make jagged edges, while others don't seem to do that so much.
 
Yes, it's artifact, no moiré. You can't have moiré with Foveon technology.
Sure it can produce moiré.
What you mean is color-moiré.
Can you explain please?

For me moiré is caused by interpolation used by a Bayer sensor. Since Foveon doesn't interpolate anything because it has "every pixel information", I don't understand why it could produce moiré... (but I'm not a pro I have to admit...)
 

Keyboard shortcuts

Back
Top