Strange artefact in water

[jimbobuk]

As described in another thread i took my R-D1 for a walk about yesterday. I'm mostly pretty happy with the shots it took and the handling of it.. more practice is needed and more looking over what i got to pick out some favourites to get online.

Anyways i did notice in this one shot that the water looked a bit strange. I've attached a crop of it, which at normal viewing is a bit subtle, if you could download it and load it in your photoshop equivelant and then magnify it 2x you can see the effect much more pronounced. Strange vertical lines in the water that to me resemble interlacing artefacts you sometimes get in videos. Its present directly from raw to tiff, so its not a jpeg artefact, tho it'll probably have some of those on this attached file i send to you. I've tried a couple of raw programs and it always seems to be visible.

Its not THAT visible at 100%, just a curious thing really, obviously the water was moving a little but its a reasonably fast shutter speed to capture the birds mid flight.

Any ideas?

 

[Gid]

Jim,

I can see the effect you describe on the attached file at 200%, but as the file is only 55Kb its difficult to pass comment. Is this visible on the original raw file?

Regards

Gid

 

[jimbobuk]

yes the jpeg hasn't added too much at all.. I've noticed that the files sometimes don't look great at 200%.. i've been applying too much sharpening in raw conversion i think for most of those occurences, usually people have said how well the images from the R-D1 enlarge for their 6megapixel size..

i'll get another pic of it online in lossless format.

http://jimtreats.com/epson_R-D1/problems/EPSN1080_crop.png

 

[Gid]

Jim,

Try re-doing the raw file in Epson raw with everything set to default and see if you get the same results. I always leave everything at default except for white balance if adjustment is needed. I've never seen anything strange in any of my images (other than what was meant to be there 🙂 ) If I need to sharpen I do that in photoshop. I haven't had any issues with moire, which I am beginning to think might be what you are seeing here. However, there are more experienced guys here that might be able to shed some light on this.

Regards

Gid

 

[Doctor Zero]

Oh no! This is scary! The prophecy is coming true: "polywater" does exist after all! The Russians were right! We are doomed!

Sorry, I couldn't resist. I hope I'm not the only one who thought of this... anyone... hello...?

I'm afraid I've nothing constructive to suggest other than what's already been suggested. It does look as if it could be a compression artefact. The ripples are fairly regularly spaced, which may interfere with the compression algorithm (I think they use wavelength/frequency analysis of an image).

Doctor Zero

 

[jimbobuk]

I've gone through more shots for other purposes and came across another shot that shows it.. most of my water shots at that same place exhibit that problem when the ripples get small... this shot though is a fine branch of some trees.. the image is rotated so the vertical glitches are horizontal in this one.

Its not terrible its just that it IS noticable at 100% viewing if you look close.

It seems that lines being thin and at 45 degrees exagerates it.. which makes sense this is where the stepping would be most evident as you get down to the pixel level.. it could be some interference with the anti-moire/banding filter the sensor has?

Anyone else see this in any of their images knowing to look for something similar?

I'll try the epson plugin when i can, not on my current machine at the moment.

 

[kenpex]

The problem is in the RGBG array (ok, I'm not sure, but it seems a quite classic artifact of one of the algorithms that is known in the literature for that task) -> full colour pixels conversion (demosaiking) that the camera has to do to recover full colour information from the sensor. If you shoot in raw, then using a better raw converter should solve this problem. If it does, then I'm right 😀 The fact that it happens more at 45 degrees lines is consistent with that, as the problem is that the algorithm I'm talking about chooses locally a direction (vertical or horiziontal) to interpolate information, and if it guesses wrong, you'll get that artifact.

 

[jimbobuk]

these shots are taken from a more aggressively sharpened bunch of images, but even with the sharpening reduced its still visible just slightly less pronounced.. ie. i think i've over sharpened for 100% viewing, not for <50% sharpening though perhaps.

They were made by capture one pro, which i thought was one of the better raw converters available.. will try in epson and raw shooter when i can.

Thanks for the input though.. it makes sense to me why something like it would happen.. be nice to know a few other R-D1s did it so its not some problem with the camera, though i know its not going to be really.. such a specific thing.

 

[pfogle]

Well, my 2c worth (and I have to say I'm guessing) is that it's aliasing, though I don't know of any other reports on the R-D1. It's certainly more apparent on the Nikon cameras, as opposed to Canon, which I am told is because of the weak anti-aliasing filter on the Nikon chips.

As the R-D1 uses the same chip as the D100, perhaps it's the same problem? If so, I don't think there's much you can do, except use RAW and do all the sharpening after the conversion.

ps for Dr Zero - I orinally thought this was a thread about homeopathy 🙂

 

[Trius]

OMG, I had TOTALLY forgotten about polywater!!! Looks like the Soviets will win in the end after all!

 

[RML]

Weird. I've never seen any such artifacts on any of my shots with the R-D1. That said, I don't use any in-camera sharpening, contrast or whathaveyou at all. If I need them I'll do it when I'm developing the RAWs in the Epson RAW converter.

 

[jimbobuk]

Ok, I finally got round to trying out different raw converters, I've made a composite image of the two shots reconverted with the 3 programs.

http://jimtreats.com/epson_R-D1/problems/R-D1_Artefacts.png

The Capture one capture is with 0 sharpening as opposed to near 200 sharpening that was on the original shot.

The epson converter does a much better job.. the water picture though still aint perfect.. much less obvious at 100%, but at 200% its bizarely blocky. Anyone seen anything like that with their shots?

The trees shot is handled much better, not sure if the tree had that patterning on it but its not offensive at all..

Its clear as i've read before that rawshooter is sharpening quite a lot even at default settings.. most C1 shots aren't any better with no sharpening, so its clearly an issue with how they're getting the data out of the raw file.

Its also clear that the epson as shot isn't very sharpened at all. this could be helping mask the problem.