Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

All the Epic/Dragon Vs. Alexa/Sony threads are belong to us...

I plan to post some Dragon samples without OLPF.
Some tests I did the other day with normal Dragon vs Epic without OLPF and my processing showed that I have been out resolving Dragon lately.
And this is without Super resolution methods involved.

Dragon OLPF is very tight optically, which is good. I see minor false color debayer errors on resolution charts. Normal scenes have no issues at all.

Gunleik Groven said:
it is interesting to see the difference in performance under test/chart conditions and practical images for the F55.
trying to find out if we shot WO OLPF. but even if we did, it seems kinda steep.

it os a consistent tendency in all the images i have, though.

the alexa does "as expected", really.

there was a surprise cam doing a lot better and worse than expected, but the results are so erraric, i have decided to not use them.

g
Click to expand...
 
THX Gavin!!!

Sorry.
The flower example was also faulty.

There was a reason I wanted to put that in...

Updated now:

http://blog.gunleik.com/?p=1263

d4f0a781a8feecf576448d1512bf5aa1c56d686d.jpg
 
Last edited:
On the topic of "classic debayer" vs ADD
Looking at the REDcon A.D.D thread images:

On the plus side, ADD dramatically reduces the noise floor and removes black crush.
ADD3.jpg


But I also noticed quite a lot of "color ghosting" in moving images, particularly in the hair and clothing of this skater
(Right click -> Open in new window)
ADD2.jpg


I don't know how they use previous/upcoming frames, but maybe it's a failure case in their temporal re-projection algorithm.

Just my 2c,
Matt
 
Can some one from this forum please do and post a side by side test of red dragon sensor vs. 35mm film. Would love to see the dynamic range and colors pitted gainst the gold standard that "FILM" is...

As the destiny of digital image capture and ultimate goal is to beat film and exceed it in terms of overall image quality, it would be great to see this sensor, which was hyped and marketted as "THE" sensor that will have more DR than film and will retire it, in a side by side test with film.....
 
Also would be great if you can throw alexa into the mix....:redface:
 
Matthias Hutter said:
But I also noticed quite a lot of "color ghosting" in moving images, particularly in the hair and clothing of this skater
Click to expand...

What you describe sounds like what Les was talking about as well. Hopefully it is just a bug that can be fixed.

Here are two images from Les, the first w/ ADD and the second w/o. You can see the color artifacts in the branches of the trees. If you are on Windows, you can use the zoom button in the bottom of the browser to see it better. Sorry for the image size differences, that is what Les posted.

With ADD
DRAGON_OLPF_ADD.024.jpg


W/O ADD
DrAGON_OLPF_RCX.024.jpg
 
Stacey Spears said:
What you describe sounds like what Les was talking about as well. Hopefully it is just a bug that can be fixed.

Here are two images from Les, the first w/ ADD and the second w/o. You can see the color artifacts in the branches of the trees.
Click to expand...
Yeah, this the same issue, albeit maybe with a different cause - occlusion/disocclusion of tree branches.

As for "fixable", it depends on their implementation, which is obviously proprietary so I can't say for sure.
I know of a few techniques to use information from multiple frames e.g. temporal reprojection, mesh-based warping, spatio-temporal anti-aliasing etc.
Generally the algorithms that ghost the least are those that focus on the actual difference between frames, e.g. with the accumulation buffer.
 
Converted the TIFF in Resolve to Rec709 with Arri's LogC to 709 LUT. The Alexa shot is noisier in the low end and to me, has far far far less detail, and seems to fair a tad bit worse DR wise in this comparison. If we were to bring down the Alexa low end to match noise levels the difference would be more obvious that Dragon excels here. Framing is different so can't be too definitive. Anyways, check out the files and make your own conclusions. Thanks again Kemalettin!

Alexa:
AlexaLUT.jpg


Dragon:
Dragon.jpg


Alexa brought down to match Dragon noise level (Dragon is still better noise wise):
AlexaNoiseMatch.jpg
 
You must log in or register to reply here.
Back
Top