Telecine vs. Coloring in Avid

[Marc Levy]

If not going back to film, does coloring in Avid yield the same results as coloring in a telecine session? I just colored some stuff, and it looked great on the small monitor, but when projected, there seemed to be a digital grain. Was this added "digital grain" caused by color manipulation in Avid? Would this digital grain appear if the same color effect was produced in a telecine seesion?

 

Should I pay the extra money to supervise a telecine session, or should I just color on Avid?

 

Thanks

[Phil Rhodes]

>If not going back to film, does coloring in Avid yield the same results as coloring

>in a telecine session?

 

No.

 

The negative will always contain the largest amount of data; if you print it to more film, the print will be higher contrast - information lost from the hilights and shadows. However, if you transfer it to video, this effect is even more pronounced - really you're throwing away most of it by going to tape. The net effect of this is that when colouring in telecine, there's more information there than what's on the monitor - there's often somewhere else to go in light or shade which isn't visible until you start tweaking. On tape this isn't so - what you see is what there is, and you can only bend what's already visible.

 

And secondly, the problem you noticed - compression. You don't mention what Avid system or compression setup you're using on it, but most tape formats (from Digibeta down) apply some kind of image compression. These algorithms throw away even more data, usually in such a way that it isn't visible or isn't very visible in the circumstances. However, if you change those circumstances (particularly by increasing contrast) the artifacts can become more visible.

 

However this isn't to say that there's no place at all for electronic grading. It's certainly enormously cheaper than doing it at transfer, because you only have a £10,000 piece of equipment waiting on you, rather than a £2,000,000 telecine, and those of us who're stuck shooting tape have no other option. It really depends on how picky you're being, and how extreme a colour effect you have in mind. Small tweaks on video are usually fine - even to the extent of fixing grotesquely out-of-whack colour balance, even on DV source material.

 

However personally I think that there's no point shooting highly expensive filmstock to cheap out on the postproduction.

 

Phil

[Marc Levy]

So, tell me if this is what you're saying. If I plan to increase contrast and saturation to a large degree, I would yield better results doing it in telecine than in AVID DV EXPRESS, even if tape is the outcome either way?

[David Mullen ASC]

The color-corrector attached to the telecine is bound to be more sophisticated than your AVID DV Express software color-corrector, but also you'd be working with a highly compressed recording with a lot of color-subsampling compared to working off of the negative with all of that information.

 

On the other hand, adding contrast is really about throwing away information (clipping the highlights, crushing the shadows) so doing it yourself might yield OK results. But adding saturation is harder to pull off if working with a DV recording -- there's less color information to manipulate so you tend to get noise as you boost it. In that case, you'd be better off doing it in the telecine transfer session.

[Alvin Pingol]

But adding saturation is harder to pull off if working with a DV recording -- there's less color information to manipulate so you tend to get noise as you boost it.

I've always known this as fact ever since I first started manipulating DV images, but I have never really understood "why."

 

When you look at a raw DV image, it doesn't really look that bad. You can't see color artifacts unless you look for them. In fact, the image, color-quality-wise, looks similar to other images that have high color bit-depths.

 

However, as we all know, these DV images get nasty really fast as soon as any form of color manipulation is applied. An increase in saturation will immediately reveal these problems.

 

What is the technical reason behind this? Is it due to DV's low color bit depth? Compression? YUV?

 

I do know the whole deal about GIGO, but would converting a DV signal to a higher color depth and/or RGB (vs. YUV) help at all with color manipulation?

 

It just really upsets me when I look at a nice-looking DV image, then try to slightly manipulate its colors only to bring up nasty artifacts...

 

 

EDIT: Mr. Mullen... "Advanced Member," eh? Niiiiice! B)

[David Mullen ASC]

I have the same problems color-correcting HDCAM, which is 3:1:1 (which is better than 4:1:1 but worse than 4:2:2) and highly compressed.

 

Looks great but as soon as you start to push the color around dramatically, it's noise and artifact-city.

 

Yes, it's due to both having too-little red & blue information plus a lot of compression.

[Phil Rhodes]

Hi,

 

At a mathematical level I get the impression it's to do with how the DCT components are weighted when creating a macroblock. The compressors, briefly put, set up each block to reproduce an optimum image under the circumstances in which they are compressed - they know what the contrast is, what they can get away with, and what the results will be. The compressor intrinsically compresses an image based on what it is now - not what it might be after you've pushed it around.

 

If you increase contrast or saturation (which is really also an increase of contrast, just between RGB channels) you can push these compression artifacts out of invisibility. It's completely intuitive really, you just have to consider the way the codecs work. This applies to everything from DV to HDCAM (HDCAM is extremely similar, just "more of it.")

 

Phil

[Alvin Pingol]

Ahh, I see. Thanks.