Strong artifacts

Started by ryan750mhz, November 18, 2017, 04:44:38 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

ryan750mhz

Just testing RAW on 50D, loaded the CDNG at resolve and got this very bad artifacts (strong green/purple high contrast lines) on my image, is this beyond "normal"?

came from Oct042017 nightly for 50D >> converted to Cdng via raw2cdng 1.7.9 >> Resolv

This is a Jpg export From Lightroom (adjusted WB)





This is a jpg export from Resolv



Walter Schulz

Without knowing about ML build version, settings, postprocessing tools: Yes. No. Maybe.

ryan750mhz

@Walter
sorry my bad, edited my post with details already

reddeercity

That's what you call aliasing & moiré problem -- the result of line skipping and pixel binding . Just search for aliasing & moiré patterns .
Take the 50D sensor size in photo mode 4752 x 3168 divide by 3 gets the camera resolution in 1:1 1584x1056 for video
But In 3X crop mode where every pixel is counted (no line skipping or pixel binding or at least minimal pixel binding)
So the lesson here is do not shoot -- bricks , power lines , wires anything that has fine patterns in 1:1 mode
but it's ok to shot that in 3X Crop mode where the problem dose not happen .

a1ex

Looks pretty strong; do you mind uploading a short MLV (1 second is more than enough) for this experiment ?


a1ex

Thanks; using the previous and next frame gives some minor improvement, but nothing worth showing.

However, I'm not getting your artifacts when rendering the image normally (dcraw -a -b 2), and I'm not even using an alias-friendly demosaic algorithm.

Before (original, uncorrected): M18-1635_000003-dcraw.jpg
Corrected with 1 frame before/after: M18-1635_000003-dcraw-a11.jpg
Corrected with 3 frames before/after: M18-1635_000003-dcraw-a33.jpg

edit: also tried dcraw-float; interpolation type 6 (LMMSE) does a good job reducing color artifacts without any additional trickery; I'd expect the same performance for any GUI raw processor that uses LMMSE (e.g. RawTherapee).

Same example with RawTherapee, LMMSE:

Before (original, uncorrected): M18-1635_000003-RT-LMMSE.jpg
Corrected with 1 frame before/after: M18-1635_000003-RT-LMMSE-a11.jpg
Corrected with 3 frames before/after: M18-1635_000003-RT-LMMSE-a33.jpg

ilia3101

Tried the MLV out in MLV App, with AMaZE demosaic, but the results looked worse than LMMSE:



@a1ex It seems to produce a result not much better than the 'not even alias-friendly demosaic algorithm' you used. And definitely much worse than LMMSE.
I'm sorry for hijacking the thread, but do you possibly know why AMaZE results in so much colour aliasing in MLV App?
I compared it with AMaZE in MLRV, and MLRV's AMaZE produced much less alising: https://github.com/ilia3101/MLV-App/issues/47

a1ex

Unable to answer without sitting down to debug it, but from what I remember since I've used AMaZE in cr2hdr, it seemed to prefer white balanced data as input (so it could know what's grayscale and handle aliasing better in these areas). At least that was my understanding (or maybe my implementation is also broken). I don't really understand how it works (tried to ask somebody who seemed to know), only figured out how to call it.

ilia3101

Ah yes, thanks for the white balance idea. I really must try that out.
And wow, that somebody really does make AMaZE sound particularly good (maybe its the original creator :)) )