Very Clean Andromeda - Optical White Balance

cardmaverick

cardmaverick

Well-known member
http://www.chrisnuzzaco.com/special_uploads/filtertest.tif

That is a cropped screen shot from AE. I'm not sure how much information gets captured during a screen shot though, so it might not be a full 16 bit file, but its much nicer than the quicktime I compressed.

Remember to view with a proper gamma of 2.2, your browser might display it wrong.

This is an 8 bit Animation file, uncompressed. Remember... its 8 bit, I shot in 10 bit, so if you try to mess with this file, you are missing a full 2 bits of data (thats a lot of extra data), and this is already been tweaked by myself anyways.

About 220 MB:

http://www.chrisnuzzaco.com/special_uploads/filter_test.mov

For those wanting to give editing the footage a whack....

Uncompressed 16 bit tiff image file sequence 436 MB:

http://www.chrisnuzzaco.com/special_uploads/filter_test.zip

These frames were captured linearly at 10 bit, and rendered linearly into a 16 bit space, so the images will look dark on your average 2.2 gamma monitor, but all the data is there. It was also rendered without sharpening. I chose to use more advanced sharpening methods in After Effects, Sculptor HD's sharpening abilities are a bit rudimentary and they tend to accent signal noise. These frames have no lateral CA compensation applied either.

A little bit of information abut the shot:

This was shot using a special manual white balance setting I created that forces the blue and red channels to equal the green channels gain. Basically its a "uni-gain" setting. This effectively turns your camera into a Viper Filmstream like system, where optical white balance brings the best results. The camera is natively balanced for daylight, so I had to use an 80A filter on top of my 90ccM filter. If you load the frame into photo shop and isolate each channel, you will see that all three channels have the exact same amount of noise, which is very low, this is due to the optical white balance. The shot was rendered unsharpened. In After Effects I broke the shot up into individual channels and compensated for the lenses lateral CA, then I applied an unsharp mask to the footage, making sure to not accentuate signal noise. Sharpening is much easier when you have footage with a very low noise floor.

It looks even better when you see it play
icon_biggrin.gif
 
Last edited:
That's very clean! Yea, let's see some videos soon!

Btw, this optical white balance procedure which you've done, would it work on normal sd and hd cameras; i.e. non-Andromedized, just plain straight to tape/media cameras?

Well what I wanna ask really is, would it do any good since the end product is gonna end up as 4:2:0 or 4:2:2, 3:1:1, etc?

Thanks
 
Thanks Guys!

This is an 8 bit Animation file, uncompressed. Remember... its 8 bit, I shot in 10 bit, so if you try to mess with this file, you are missing a full 2 bits of data (thats a lot of extra data), and this is already been tweaked by myself anyways.

About 220 MB:

http://www.chrisnuzzaco.com/special_uploads/filter_test.mov

For those wanting to give editing the footage a whack....

Uncompressed 16 bit tiff image file sequence 463 MB:

http://www.chrisnuzzaco.com/special_uploads/filter_test.zip

These frames were captured linearly at 10 bit, and rendered linearly into a 16 bit space, so the images will look dark on your average 2.2 gamma monitor, but all the data is there. It was also rendered without sharpening. I chose to use more advanced sharpening methods in After Effects, Sculptor HD's sharpening abilities are a bit rudimentary and they tend to accent signal noise. These frames have no lateral CA compensation applied either.
 
Last edited:
Luzer said:
That's very clean! Yea, let's see some videos soon!

Btw, this optical white balance procedure which you've done, would it work on normal sd and hd cameras; i.e. non-Andromedized, just plain straight to tape/media cameras?

Well what I wanna ask really is, would it do any good since the end product is gonna end up as 4:2:0 or 4:2:2, 3:1:1, etc?

Thanks
Click to expand...

I haven't done enough testing with this yet while shooting DV, so I can't really say, but I think you would still benefit because you are creating a cleaner more optimal signal right off the bat prior to compression. I have noticed that I DO get more dynamic range when viewing the DV feed while shooting.
 
Wonderful frame! I wish the possibility was there to make PAL Andromeda.

Thanks for the frame!
 
muonwave said:
90ccM filter? Is that the series of magenta filters described in another post of low-noise you'd written?
Click to expand...

Its not the same density of filter. Its the same color, magenta. Thats an old post, before I figured out the proper density required for full optical correction. At the time I was doing only partial correction with the 60ccM filter. I'm technically using three 30 ccM filters. That shot actually used 5 filters! A UV filter, three 30ccM filters, and one 80A filter. You don't need the 80A when shooting under daylight though.
 
William_Robinette said:
Thats crazy clean. Nicely done.

How many stops of light loss are we talking here?
Click to expand...

Umm... thats a good question. I just measured the setups...

3,200 Kelvin Filtering:
90 cc M + 80A: -4 stops

5,500 Kelvin Filtering:
90 cc M : -2 stops

The 90 cc M on its own is only 2 stops, so its like an ND filter, which is nice because you usually need to use ND filters when shooting outside anyways.

As for the heavy filtering required for indoors... lets just say super clean images don't come free of cost. The Red One has the exact same issue as my camera, you need to filter and use more light, or just shoot with HMIs to get the best image out of the camera.
 
Chris - you mentioned something on the reel-stream forum in an earlier post about getting 12-bit firmware upgrade. What was that for, and what firmware were you talking about?
 
Last edited:
Ah yeah, the 12 bit firmware upgrade I pushed so hard for...

Currently the system is 12 bit to 10 bit via LUTs. Juan told me he had created a firmware upgrade that allowed 12 bit linear (eliminating the need for LUTs). It sounded like you would need a raid to record it though, but thats not very hard to come by these days, so it was feasible. The last I heard, which was months ago, was that he still needed to run some tests on it to guarantee functionality.

I can't tell you how tempting it is to sit down and pull off the bottom plate from my camera and take a look at the inside of my camera.
 
I hope he'll still do that. Though, unless it's an uploadable solution (re-writing the existing eeprom - assuming there is one in his mod - the LUTs have to load into something), I'd be curious how he'd get it out to people. I'm not sure about sending the camera back for a re-fit (I'd send my other one, and ask for the full install) - not when I'm on a schedule for a feature.

I've also "almost" took it apart several times with the intention of re-engineering/reverse-engineering the mod. One of the things I was thinking about to add to it was the ability to balance each channel separately (creating the "uni-gain" WB you've been engineering outside of the camera). Sort of a limiting circuit with several presets (think of each preset as a programable LUT for that channel able to be retrieved and applied with the press of a button) - I'd love to see his schematics on this. Probably shouldn't say any more....
 
You must log in or register to reply here.
Back
Top