Tragic Lantern for 6D

[xNiNELiVES]

I'm also kind of lost

Ok good to know that I'm not alone lol. I was kinda embarrassed.

[teo770]

Understood, with last update, when clicking "don't click me", a  bunch of raw tests begin
So it generates DNG files of different type.... (resolution ?) to let us check what is the best moireless type, right ?

[xNiNELiVES]

Understood, with last update, when clicking "don't click me", a  bunch of raw tests begin
So it generates DNG files of different type.... (resolution ?) to let us check what is the best moireless type, right ?

Is this to find the next highest resolution?

[1%]

Remember af_raw and the pixel defects? 5D3 had a bin that let you try all of them and then they picked one to use... my cineform conversions all have yellow and purple pixels around sharp edges. they do not on 50D footage.

[xNiNELiVES]

Remember af_raw and the pixel defects? 5D3 had a bin that let you try all of them and then they picked one to use... my cineform conversions all have yellow and purple pixels around sharp edges. they do not on 50D footage.

Ok so does it save a custom profile on the camera? Or do you want us to send the logs to you, where you'll then analyze the results and choose the best setting?

[teo770]

So we get 100 DNG files.

71 of them are breaked => trash
29 of them seems to have different expo/noise level

The goal is to find which one is the best ?
Less noise ? Best details ?

for myself, it's the #25
Sharpest, lowest noise, bit of vertical banding in blacks

So... we can modify a setting according to this choice ?

[1%]

Pretty much. Looking to not have any edge damage like this, then I'll add it to the next build.

http://s14.postimg.org/t6vuw7std/M30_1221_proxy_avi_snapshot_00_02_2013_06_02_12.jpg

[teo770]

In which order is it important ?

#1 - the less moiré (i presume)
#2 - the less banding
#3 - the less color noise
#4 - the less b/w noise

Will this pic choice be the same for all 6D cameras ?

[1%]

Should be the same for all 6Ds.

Remember its n-1 too for type vs image so specify image or rawtype


Probably shouldn't affect morie but will affect #2 + 3, maybe 4.
See what is easiest to correct... these color outlines are hard and ruin cineform workflow. I dunno what the banding will do or how hard it is to get rid of, etc.

[noisyboy]

Hmmm... 25 was a trashed file for me. 35 and 73 were best/least noisy at my end.

I'll run the test a couple more times tomorrow. Just a bit late now

[noisyboy]

Pretty much. Looking to not have any edge damage like this, then I'll add it to the next build.

http://s14.postimg.org/t6vuw7std/M30_1221_proxy_avi_snapshot_00_02_2013_06_02_12.jpg

Sweet! Didn't realise you were trying to combat this! Thought this was just something we were stuck with :p

Will definitely run a couple more times when I get up and make sure to shoot something that might give use that problem.

[oscaroo]

Thank you 1% and friends.
I deployed "[6D] RawtypeTEST.zip" on one memory card and found:
 - format works
 - audio tags work
 - wifi works
 - file browser works
 - raw zebras work in LV but not review (I assume by design)
 - in camera raw processing works.
 - focus peaking works

Everything I ever wanted works! YAY.

[xNiNELiVES]

Well I still have to drop the money to have may 11-16 fixed.. I get the feeling it will be a winner at 11mm since now we crop so hard, that vignette will just go away.

On FF the 11mm is like what 8mm? 6mm?

Not to be annoying but, when or are you going to unlock the width to 1880? I don't see the purpose of the child locks.

[teo770]

With update: [6D] RawtypeTEST.zip
Sandisk Extreme Pro (95MB/s) formated exFAT - Exp Override ON - Global draw OFF - FPS Override 23.976 - Preview Canon
Debug: dialog timers disabled

1600x640@23.976p: 12.15 minutes / 30.31GB without frame drop ! => Card full (32GB)
1600x670@23.976p: 1.24 minute without frame drop
1600x680@23.976p: 1.12 minute without frame drop
1664x624@23.976p: 3.50 minutes without frame drop
1664x666@23.976p: 55 seconds without frame drop
1728x648@23.976p: 50 seconds without frame drop
1728x692@23.976p: 30 seconds without frame drop
1792x598@23.976p: 1.27 minute without frame drop EXTREME !

[noisyboy]

Pretty much. Looking to not have any edge damage like this, then I'll add it to the next build.

http://s14.postimg.org/t6vuw7std/M30_1221_proxy_avi_snapshot_00_02_2013_06_02_12.jpg

In fact - taking a closer look at that image - I thought you meant the discoloration (which I guess is moire?) but I have never seen the horizontal dots in anything I've shot before...

[teo770]

(which I guess is moire?) but I have never seen the horizontal dots in anything I've shot before...

This is a kind of moiré/artefact affecting edges.

"Real moiré", you can't miss it, just shoot something with a lot of vert/horz. lines.... you will find it easily...
You will also easily see aliasing...

Here, my favorite target is my slate roof.... I always know where to find some moiré & aliasing ....

In the 100 dng files.... quasi no difference concerning "real moiré"...

[noisyboy]

This is a kind of moiré/artefact affecting edges.

"Real moiré", you can't miss it, just shoot something with a lot of vert/horz. lines.... you will find it easily...
You will also easily see aliasing...

Oh yeah I know I just never saw the dots before. Is it actually a problem on the 6D?

[teo770]

I think while RAW video doesn't path thru Canon optimisations... it's part of raw data.

Testing those 100 dngs, i presume it's in fact testing diferent timer options to find
the best producing the less artefacts, no ? (1% ?)

Getting raw data, you also get sensor/electronic weirdness like color shift, artefacts....
Once you know them, some can be fixed while recording file in cam or while converting DNG
with raw2dng.

[1%]

Its testing different raw types like 5d3 to try to get rid of any dots like that, banding, etc.

[noisyboy]

Its testing different raw types like 5d3 to try to get rid of any dots like that, banding, etc.

I'll be honest dude - I haven't noticed any banding or dots. That all disappeared for me a while back.

Just ran the tests and the only differences I noticed were exposure changes and differences in noise that are so negligible that I can barely tell which is better

I shall continue to try the tests in different environments though.

In other news!!

I think I just made some kind of breakthrough! I shot some rooftops etc to try and replicate to the orange/blue color moire artefacts to see if any DNG's from the test would yield less moire than others and sadly there was no real difference between them...

However! I just played around a bit more with the settings inside of lightroom and found that choosing 2003 Camera Calibration Process pretty much gets rid of it all together!

If anyone else could try this out in ACR or Lightroom then let me know it would help me work out if this really does help or I just got lucky on one shot

Cheers! 

[noisyboy]

Damn... Tried on some other footage and didn't work

[noisyboy]

Is this test definitely working correctly? If 25 was the best for teo770, that frame for me was completely unusable (scrambled image).

[noisyboy]

Okay - so fully aware that I'm talking to myself now but anyway

Choosing 2003 process does actually make a difference. Doesn't get rid of it entirely in the second lot of footage I tested on but definitely reduced it quite a bit. One thing I'm wondering though... when I tested this on the DNG's generated by the Don't Click Me test (which of course haven't been wrapped up into a .raw file), I get more options in ACR and Lightroom for camera calibration (difference profiles etc) which I don't get from DNG's extracted with Rawanizer. This makes me wonder if Rawanizer/Raw2DNG etc are playing a part in the artifacts?

[teo770]

Personaly, banding is present in shadow areas. Make the test in a room with low ligh, you ll find it.

Edit: .... perhaps the best banding test is "cap on"

[noisyboy]

Personaly, banding is present in shadow areas. Make the test in a room with low ligh, you ll find it.

Edit: .... perhaps the best banding test is "cap on"

Cool - will have a go