Black & White on the JVC-Incredible!

[Paul Mogg]

The other day, when I was color correcting some HD to SD downconverted material from the JVC in FCP, I desaturated the picture to black and white. The results are quite amazing, the HD resolution is very much still apparant in the SD downconverted image, and of course there is 0 chroma noise. I am now dying to try this with the HD footage itself. My feeling is that if you are thinking of making a black and white film, this might well be the hands down camera of choice if your budget doesn't stretch to film. Combine the HD resolution with the equivalent of a 4:4:4 image, which is what you get when you desaturate, and you've got one heck of an image. Please let me know if you try it yourself.

All the best

[David Newman]

I think this makes excellent sense. The camera's luma resolution is excellent, and this can be explained the camera's unique CCD sensor (using Luma,Yellow,Green,and Cyan instead of a RGB array.) The sensor will achieve higher resolution and lower luma noise than an equivalent RGB sensor. Of course this maybe the cause of the increased in chroma noise that we can see. I regularly desaturate for a black and white or sepia look and I have been amazed by the results.

[Steve Mullen]

Back in the V8/Hi8 days I found that when in India I played NTSC via RF to a PAL TV itwas only B&W. But it looked like 35mm B&W film!

I have often wondered, since I always loved B&W, if maybe we should have worked with hi-rez B&W rather than "full-of-noise, bleeding color." :)

Wow, just saw a PC Richards commercial with a HD1!

[Brian Mitchell Warshawsky]

Question:

Is it better to shoot/record in Black and White, or remove the color during post/editing?

Brian

[Steve Mullen]

As I remember, the JVC has a B&W shooting mode.

[Eric Bilodeau]

Normally it is better to shoot in black and white because the information in the transport stream (and any compression) is not separable afterwards. the chroma noise becomes part of the picture once compressed, if you de-saturate it will be much less aparent but still there( mainly because some tonalities of color near one another looking different on capture creating chroma noise are normally pretty much of the same luma level so even if color differenciation is apparent because of the number of possible variations, when you get down to grey levels, all the colors sharing the same relative luma level will look alike, that is why noise is very attenuated in this process). It has to be tested though, I do not know if the BW mode is a filter applied to the image taken or if it is a omission of the chroma in the capture process. I agree though that the de-saturation gets rid of most of the problems related to chroma noise. You have to be careful on contrast in some cases though.

[Chaim Bianco]

" Normally it is better to shoot in black and white because the information in the transport stream (and any compression) is not separable afterwards."

this is true only if your corrections are done in rgb colorspace.

if you work in yuv, then they 'seperate' just fine.

try taking an mpeg2ts framegrab into photoshop and converting from rgb to yuv("lab color"). then flick the channels on and off. youll be amazed at how the unpolluted luma channel is. (and how ghastly the chroma channels are!)


" the chroma noise becomes part of the picture once compressed"

well i guess you could say that in a 4:4:4:4 codec, that the aplha channel is part of the picture too. that fact remains, as you can prove to yourself quickly in photoshop even -- no need for a $600 hour tape-to-tape suite at all -- that if you operate on the 'uv' part in a yuv colorspace, the luminance component will go untouched.


" It has to be tested though, I do not know if the BW mode is a filter applied to the image taken or if it is a omission of the chroma"

my guess is that if you shoot with the hd10u's 'black and white' effects, it records to tape without the chroma component.

(btw, low light shooting looks really good this way.)


" You have to be careful on contrast in some cases though"

increasing or decreasing what the eye registers as Contrast should be performed only on the luminance channel. if you goose the contrast in the choma channels all your doing is affecting the saturation.

and as paulM and davidN noted, the luminance component of the hd10u's output is quite smooth (aside from the obvious clipping problem) and can withstand a fair amount of massaging.

i joked with a dp friend of mine when he asked about dynamic range of the hd10. i told him it was like b&w neg in the mids, and like color reversal at the shoulder and knee.

let's all hope for an eventual jvc merger with foveon.
.

[Chaim Bianco]

Brian Mitchell's question:
" Is it better to shoot/record in Black and White, or remove the color during post/editing?"

you can certainly extract a nice looking 'chroma noiseless' luma channel from a otherwise noisy 'color' mpeg2ts.

my question for a jvc tech would be "when recording in 'black and white effect mode" is the total bitrate being used for the luma. or is it just dropping the chroma component. my guess is the latter. although the former would much cooler.

and i guess, brian, if you wanna know the answer to your question by 5pm today, you can always just shoot a test. -- a tv sitting on a red carpet shot in low light should give give you a nice enough range of horrifying color artifacting to work with. -- at a certain point the color noise almost what appears to be a sci fi special effect.


-- btw on the topic of bitrate, has anyone considered that the panasonic varicam is *always* shooting at 60fps, whether your youre delivering 24fps or 30. -- so when they say 100mbps, at 30fps isnt it really only delivering 50mbps? or is it duplicating successive frames to tape and investing more bandwith into lower-framerate (like 24p) frames?

if that's the case, then shooting at 6p should look almost uncompressed.

[Eric Bilodeau]

Chaim Bianco -->>"well i guess you could say that in a 4:4:4:4 codec, that the aplha channel is part of the picture too. that fact remains, as you can prove to yourself quickly in photoshop even -- no need for a $600 hour tape-to-tape suite at all -- that if you operate on the 'uv' part in a yuv colorspace, the luminance component will go untouched."

As far as I am concerned, 4:4:4:4 signals are either uncompressed or compressed with a non destructive codec. I have never seen a alpha channel in a mpeg2 compression. You are right though, you can interpret the luma channel in YUV but it is still an interpretation. You don't need much testing to corroborate the fact that a luma YUV treatment is much better than a RGB desaturation i agree. The chroma channels are as bad as DV but definition makes the luma chanel very clean indeed. It is to be expected though, video compression (broadcast) never destroys the luma ratio by compressing it too much.

[Chaim Bianco]

" You are right though, you can interpret the luma channel in YUV but it is still an interpretation. "

from what i understand, mpeg is encoded in Y Cr Cb type scheme. not rgb. so really, extracting a nice clean luma is no problem.

actually what really hurts/perverts the b&w info of the hd10u is the fact that it derives its luma from a bayer-like pattern on the chip. this seems to be a fact of life for non-foveon one-chippers. and any dithering artifacts are there before the mpeg encoding. most digital still cameras suffer the same type of problem. the solution for the still cams seems to be using chips with higher and higher resolutions. its too bad the hd1 didnt scan at 2k and then downsample. i wonder what the true bottleneck in this camera is -- mpeg encoding speed, or just the cost of the parts.


" You don't need much testing to corroborate the fact that a luma YUV treatment is much better than a RGB desaturation i agree. "

extracting luma comes standard with after effects. its definitely worth the rendertime.

[Heath McKnight]

I shot my film, SKYE FALLING, in full color and just took out the color for the black and white scenes. (Used an XL-1, not an "s.")

heath