what's better for post low chroma or high?
 

There's been alot of talk here about the pro's ad cons of shooting low con and low sat for grading or trying to achieve a final look while shooting. I'd like the technical people here to weigh in on a question central to that discussion:

For post is it better to shoot low chroma and add more saturation afterwards or better to have a little too much and reduce it grading? Obviously getting it right the first time would be ideal, but if you are going to err - which direction is better?

Off hand I would guess too much is better , but I am just guessing.

If you are going to err, err on the side of not enough. There are a ton of caveats to this though.

Traditionally, formats like film and RAW (Arri, RED, Genesis, Viper, etc.) create a very "flat" image when captured. It is very easy to move this kind of footage around, change the apparent time of day, accentuate certain colors, or reduce others, etc. BUT, these formats have infinitely more information embedded in them than the highly compressed formats most of us work in.

So when we reduce chroma in the camera, we aren't really doing ourselves any huge favors if we are later just going to push those colors right back up. We are simply adding gain in the color channels and that's not all that helpful.

However, this situation is FAR preferable to color clipping. It's just like recording an image that is too dark. We can push up the gain and get the image as bright as we want, but we bring the noise with us. However, if we clip the image then there's no saving it. Color works the same way. If we oversaturate the colors that is not fixable in post. So it's preferable to have not quite enough color than too much.

I tend to shoot a lot of my footage very flat. I realize that shooting that way really is going to introduce some color noise. I am willing to live with that for the flexibility of coloring in post the way I want. Not everyone would be willing to do that. So you take the trade-offs either way.

just like the Lumenescence , if you shoot to low , you can crank it up, but if you shoot to high its clipped off the top and gone forever.
and when they shoot to low, i hate it, like say 50% IRE, because cranking it up the range is gone, if i had 256 possible locations for the luminescence mr. to low cameraperson now cut me off to 128 steps already. magically recreating the other 128 doesnt happen.
so IMO you dont shoot lumenescence to Low OR to high.
i dont want a post gained up mess anymore than i want excessive white clipping.

with color
what does the camera originally deliver (default) Because SOMEWHERE i will be processing it , before compression codec or after? i think i would prefer before
i dont want it to high and not having a range because its (say) only 128 red steps because of oversaturation.
i dont want it to Low, because if it isnt THERE i cant magically recreate it.

i want to SEE the color HOT HOT HOT, i crank up the viewer at Least, so i can see any offset white balance or colorations , because fixing that stuff in post doesnt work like it should/could , if the colors are screwed up its a mess to get it back, if at all.
if the saturation is so low i am not seeing any of it's major tint problems, or color offsets then i went off blind , and made an irreperable mess.

i dont want to control it all the time in the Feild, especially if i am going to have to process it anyways. the freaking camera should be able to deliver a consistant proper saturation, neither to high or to low, with the human setting temperatures , white balance and controlling lightings and such.

if i dont have to post process it from totally correct original, then i might not have to de-re-compress the signal At ALL ever (in my dreams), except to put it on the output format, and the storage format is also cleaner less de-re-compressed image.

i want it all perfect , the first time , and i expect the camera to at least assist in the adventure. do it right, or go home :-)
Are these thing that hard to control now? minus the 500 menu settings to reek havoc on the internal processing?

Thanks Perone,

You might be talking about a more extreme situartion than I was thinking about though.

How far does your color need to go to create "color clipping". I'm not talking about and extremely saturated picture, just something perhaps a bit rich that you might turn down a touch in post. I realize that anything way over-saturated would be destructive. For example I don't think shooting in say std 3 with the matrixes all at 0 and using High Sat tends to a extremely over-saturated picture that would be clipping would it?

It sounds like you are acknowledging that adding color to a flat picture does add a bit of noise. Aside from clipping is there any other negative to shooting a "rich" picture and turning down later?

Without testing - so I'm guessing here - it would seem to me that you would want to have enough information from each of the colors so that if you want to alter it there will be something there to pull up.

I've always noticed that when chroma is low - either in the monitor or the camera itself, when in situations where you can adjust lighting or can cheat white balance - there is a common tendency to want to warm up what seems like a dull image and thus end up with something that is pushed way too much to the CTO side of things. I think that's what Marty was talking about.

A lot depends on the camera. 8-bit color (at best) isn't a whole lot of space. Point your handycam at the sky on a cloudless day, and you'll get to see the colors break.

I liken the idea of grading from a flat image vs grading from a well saturated one as akin to coming to a fork in the road. If we choose a direction, then check the map 1 minute later, we can verify we are on the right road. If we chose incorrectly, it's easy to back up and go the other way. If we chose correctly, then all is well.

Let's say we go to shoot a scene for a movie. We are out in a hayfield and it's 10am. Bright sunny day, not a cloud in the sy. We finish all but a few lines because the DP says we've lost the sun. We decide to complete the scene the next day. The next moring is overcast, flat, and gray. How do we seamlessly combine the two days?

If we shot day one with full, rich color and strong contrast, you're done. If you shot it fairly low contrast and somewhat flat, you can get a reasonable match.

I am catching hell right now trying to do a color grade on a short film, where the lighting was changed on an indoor scene, and it was shot to look good out of the camera, even though everyone KNEW we were going to do an extensive post. I am trying to invent lights in post, get shadows to match, get skin tones to match etc. Biggest pain in the neck in the world.

So in "perfect world", yes, shoot everything with rich, bright colors and nice contrast. Nothing will ever go wrong, footage will never get corrupt, there will never be a need for a reshoot at a different but similar location on a similar but different day. In the REAL world, this kind of thing happens all the time.

thats good
i could compare that to Scanning Pictures with a scanner.
The scanner software has all sorts of processing TRICKS that it can do to the scanned image, they often work ok , but if you really want the RAW unretouched scan that comes off the scanner imager, you have to turn all that junk OFF.

you get this flatter, less contrasty, less sharpened, uncorrected image thing that looks pretty bad , dull, lifeless.

load that up into Photoshop and you can do anything with it, no attempt yet has been made to do the auto corrections stuff.

because that is ALL the imager was giving the software anyway.

The software would try doing all its cheap tricks in its semi-automatic mode, and it isnt as controlled for EACH picture as well as, the human could do to it manually.

the only difference here is we already are going to turn it into a low bit rate compressed picture for storage. :-) THEN load it back in for processing.

Thanks Perrone,

That's very interesting and the opposite of what I expected. I guess I'll have to do some tests of my own.

If you have any images that illustrate what you're talking about I'd love to see them.

YouTube - Fountain_cut1_Ungraded_SD.mp4

YouTube - Fountain Cut1 Graded HD

I've done some other examples, but don't have any handy at the moment.

wow that must have cost a few bucks to use Billy's music?

We're tight like that...

It was supposed to go up and come off. I really need to replace that music... in fact, I'll do it now. Thanks.

Perrone,
To be honest that sample doesn't convince me as to me it looks like what I might guess would happen from a low sat original.
I don't mean to criticize how you colorized the shot - that's fine and it belongs inside a sequence of some kind anyway.
But I do notice that individual colors don't have a lot of their own contrasting saturation.

It looks like to get saturation back into the red stones you've had to add warmth into everything including the whites of the water and especially the sky and the lamp. Of course its a warm late afternoon look , but it looks like the original had a slightly blue sky. Even the foreground blues look like they've lost some of the purer blue quality ( did you use a warm grad)

Its moody and probably what you wanted, but doesn't have a lot of pop to the color. Could you have added color in a way that would have created more pop ( without taking the time for secondary color corrections?

My guess is that had the original more saturation you could have gotten maybe the same mood, though perhaps the mood would have been harder.

Actually it looks like an extreme example - do you always shoot with so little saturation.

On a related note isn't it easier to grab colors for a secondary color correction if you already have more saturation.

It was an extreme example. I'll have to work on getting you something more "normal". Give me a few minutes to work something up.

Let's try this instead.

The original image is not mine. It was downloaded from Cinematography.net as a RAW Viper image as a DPX.

What we have are the original, and 3 processes of work that I did to it.

1. Original
2. Contrast correction with 1-light color pass (aka white balance)
3. Color correction with 3-wheel color corrector
4. Saturation added, color tweaked, output set of computer RGB for showing online.

From here, someone would then do a color grade to match other footage or impart a "look". The idea here was simply to get back to "normal". And since I have no color reference I am absolutely guessing at what the scene might have looked like.

Better yes,
But it feels to me I'm still not seeing much color in the city or the river. Still doesn't pop to me. That might be just the effects of haze, and it might be appropriate for the shot, but could it also be the result of a low chroma original?

Pop? Dude they're a half-a-mile up! Of course it's not going to pop. Have you kooked at a photo of London even on the ground?!

I've got another one coming with more color...

Ok, another reworked DPX file from a Viper

Same deal...

1. Original
2. White balance with contrast correction
3. 3-wheel color + computer RGB translation

Well that's poppin' more.

I need to do some tests myself though to really see when you start reach the limits on either side of the too high or too low equation.

I've been emailing a colorist friend who said with lower bit rate stuff like an EX-1 he suggests getting as close as possible to the final look in shooting.


BTW - Yes I did understand that they were probably a half mile up

Thanks Perrone you are doing alot of work to post all this and I appreciate it.

What on earth is that thing in London? Looks amazing.

On my crummy LCD, the uncorrected youtube video actually looks better. The blue is bluer, the back more black, and the water looks more refreshing. In the corrected version, the bg looks orangey. Just one man's puter.

Thanks for the the posts Perron.

The London Eye's official website for the best online ticket prices, guaranteed

Man, calibrate that monitor! But yes, the "graded" version was done as an example to show some things, not as a "look how great this is" piece. I've just left it there as a demo.

ha ha, anybody shooting an original like that pic 1of4
http://www.dvinfo.net/forum/attachme...doneye_raw.png

doesnt need grading, they got an F, and should go get a real job :-)
what was someone thinking to shoot out of thier camera like that? did they have a color viewer anywhere on the premisis, do they know what the letters WB stand for?

although i admit it makes a perfect Alien Landscape, with the cool contraptions in the pic and all.
Here we are on the planet Krendar the green haze caused by the methonol chloride gasses the locals breathe :-) as we overlook the city in our cloaked observation booth, we pray that the seals hold out the deadly green toxic gasses. Our research has shown that the krendar Retina compensates for the gastly hues on thier planet, so as we observe, we realise that a real Krendarian would be seeing a normal picture, properly exposed with the colors correct.

Well Marty,

That image was shot with the Thompson Grass Valley Viper. You can read all about it here:

Viper FilmStream Camera (LDK 7500) | Grass Valley

It's the same camera that was used to shoot "The Curious Case of Benjamin Button". Maybe you've heard of it.

The unique nature of the RAW capture of the Viper leads to the green hue that you see. This is normal and expected. And it happens in nearly all cameras because they tend to sample the green channel twice as often as red or blue. So if you are able to capture the image with ZERO processing, the images will always have a green hue.

Additionally, since the image the viper puts out is 10bit logarithmic, the images tend to look quite flat and low contrast. This is to be expected. And it takes some special processing to get that back to what you might consider normal or linear mode.

Most consumer cameras (and many pro cameras) shoot an 8 bit linear signal. Some of the more expensive cameras can get a 10bit linear signal. The camreras that do more than 10bit linear (10bit log or 14 bit linear) are usually well above $100k. The Viper is one of few that can shoot in that mode.

Most cameras shoot with what is known as 4:2:0 color. Or rather that is what their codecs can capture. Professional cameras are usually able to capture 4:2:2, which offers twice the color information. The most expensive cameras can capture 4:4:4 color but usually need external recording devices to do so. I believe the cheapest of those 4:4:4 recorders is about $60k. The actual media is extra. I believe Sony's HDCamSR recorder retails somewhere north of $80k, and the HDCamSR tapes are $120 per hour last I checked.

So I offer all that to say this:

Before you criticize the work of another, especially someone shooting with one of the most advanced digital cameras on Earth, and one that when set up to record costs more than most houses, you might want to get your facts in order.

-P

wow that camera is in a whole nother leauge.

i wasnt proposing any facts, i can (like you) understand that the imagers on this stuff probably look worse than that even, when they are raw, back behind the tricks they do to give us a picture.

i would just freak out if my camera worked that way , AT the scale of work that i do, and i would Not think it would be totally repairable.

your final repair of it still on my monitor doesnt make the city look like i would suppose it could look like with my eyes. WHICH only reminded me of times when stuff was shot here way out of wack and it was not fixable at all without laborously selecting part of the picture to process. even after the fix colors that far off still lived long after beating them down.

i mention it because after the MINOR work that we do here, i see that picture as totally destroyed. You fixed it good though, for how bad the original looks.

here in the EX forum, if you told me that an EX cam would work like that at any time i would PROPERLY use it, then i should avoid the EX cam as an option. i would not have a picture like that comming out of any camera at the level of work we do here, it is completly unacceptable.

That's just it. There ARE no tricks on this camera in this mode. Zero. No processing is done to the image at all. In this mode, all you get to adjust is focus, shutter speed, and framing.

Right. But the people shooting with this camera, in that mode, are prepared for a very expenisive and often lengthy post-production workflow.

My final repair was something I tossed together in 5 minutes on my laptop. I don't have the software to do a proper 10bit log -> 14 bit color translation, as that software costs more than my yearly salary.

The difference between working with what you are used to, and RAW images, is that ALL the information is there. Look at it this way. Uncompresed HD is about 1.5Gbps. Your EX1 records 35Mbps. You've thrown away over 95% of the signal before it hits your SxS card. (Yes I know that's exaggerated, but it's representative.) The Viper is actually recording two full streams of uncompressed HD so, 3Gbps. It has so much more data to work with than we do, it's not even worth comparing.


To me, that original is gorgeous. No clipping, no crushed blacks, full spectrum video. Gorgeous! On set, the folks who need to see an example of what it SHOULD look like when corrected, have a monitor with special software that does the correction on the fly. Alternately, the shooter could put a special magenta filter on the front of the lens that would bring things back to about normal. I created a software based version of that filter (and others) for quick correction of Viper images. It takes off the green cast rather accurately. Mine isn't perfect, but its close enough for me to drop it on, and get a very good approximation of where the image SHOULD be.

LOL! Then put a $100 Tiffen magenta on it, and it will look normal color. But it will still be low contrast. You don't HAVE to shoot the viper in that mode. You can use it as a normal video camera that produces STUNNING images without that green, low-contrast look. But at that point, you've thrown away 90% of it's quality. So people who rent it, live with it being green and get all that data they would otherwise lose...

Hey, when you're paying $15-25k a week for a camera, you want all it can give you! You can fix that green stuff later...

and just to be sure, beings my monitor isnt "grading" quality, i popped the pic into the photo program and even according to the leetel numbers for color that it shows me, the front of the buildings is a green color still.
buildings which i assume are not green originally.
so it doesnt seem to be my wakey monitor that is off.

a fuzzy select of the colors also shows that the tree colors are in extreeme similarity to the building colors , which is what i thought when i saw that. only a real master like yourself mabey would want to get handed that and told to fix it :-)
and in that case they are underpaying you and overpaying for that camera :-)

Like I said.. it's not my work. And if I had actually taken more than a couple minutes with it, I would have corrected it fully. Did you like the Chinatown finished product more? I spent a bit more time with it.

yes the 3rd chinatown picture looks great.
was that an EX camera?

Nope, that was a Viper also.

Hey Marty, you a film person?

The images you're looking at from the Viper, and what you'll get from a RED and that ilk are negatives. Few people can look at a negative and see in it an emotional and deep print - and if you do, look at the shots again and see them as a contact print at best.

In this line of cameras (and with the EX line if you're really careful), you shoot for the perfect negative, so you can do pretty amazing things when you print to positive, so to speak. The 'one light' or 'machine' prints you see in the examples are just that - a quick print to ensure the shot works, without care of 'grading', 'timing' and so on.

Yes, those terms are applied still in the world of Digital Cinematography.

Shoot flat, shoot neutral, get as much info recorded.

That's how (in one example that springs to mind) in the first Narnia film, a couple of scenes were transposed Day For Night... in POST. The digital 'negatives' (Digital Intermediate or DI) hold enough information to not only pull the shot into a particular look, but also play with virtual lighting.

The benefit is quality that survives being hammered through effects shots, compositing, grading, format conversion, compression and presentation, and the sort of flexibility that enables very complex colaborative workflows - the downside is mindboggling steps of working with digital intermediates: not just long render times, but all the brain ache about colour spaces, frame rates, storage, time code, 2K, 4K, and... I think I need a little lie-down just thinking about it.

I'd also suggest reading up about RED's workflow, and how to shoot and edit with it. At least it will put the EX in some sort of perspective.

First time I tried to grade an .r3d, I thought I was going to cry. Had an early beta of the PC software and it crashed about every other time I pressed a button! How far we've come in only a few years.

And yea, I wasn't going to go into a conversation about LUTs and color management (manglement) and all that other stuff.

I've been burned enough times with the EX1's color, that I now just shoot everything flat and tend to push a bit in post. Seems a MUCH easier way to get the color and look that I want, for not a lot of effort. I'd have to render it out anyway, so adding back a bit of saturation, and moving the luma around doesn't really cost me much time.

no not a film person (other than watching them) at all, strictally video, the only reason i am even going to the EX level is because of a few needs:
1) must handle much and much motion and full frame changes on the pic (due to crasy theatre lighting), so excessive compression like HDV will will ruin the HD.
2) must live long and proper, so having half res psudo resolution isnt good enough
3) must do very low light for stuff like funerals for example, and church stuff where they dont allow added lighting, and party stuff where the DJ insists on turning off all the lights so his light show works.

4) must have SDI out for going to switcher , and cant be HDMI because it doesnt go 100+ feet as easily
5) is shot Live
6) is displayed live while shooting in many cases
7) is displayed recorded within a few minutes of shooting
8) then finnaly is post productioned, but generally Post production is (supposed to be) limited.

9) all and or none of the above depending on the situation the job , the price, the needs of customer, and what i want to do. Meaning not all that applies for everything, but it all needs to be done with the same equiptment.

with 2-3 cams both being operated correctally 99% of the time,and a switcher that doesnt miss 1% of the time, there is little that Requires post.

There is pause and you can rollback to delete in some way in all cameras, properly used i have shot many finished products in the cam 99% correctally.

i started shooting before spending 60 hours editing afterwards even existed, so i need a camera that does stuff corectally 99% of the time, the first time. This also being Dreaming, now that everything has to go through the computer anyways.

with the pro-sumer the color had to be torqued back DOWN, which worked out pretty good.

If were not recording uncompressed fully editable full signal, then the ONLY place that full "bits" signal processing CAN occur is before it gets obliviated for these low compression rates. which is in the camera after the chips and before the compressed storage

after compression stuff is tossed out left and right. so i think for what i do, post or not , i am better off with compressing something that is already in range when the bits are still there, because after compression i cant get all the bits back it had , when it processed it in the camera.

On The Other Hand:
if the color is low and the signal is low, then THAT can BE compressed much easier, they can use thier few bits they have for the less stuff that is even there. just recovering back to all the bits that the cameras processor had wont happen unlessi fix the compression.

so i donno, i already have issues i have seen from 5-1 compression, if i was going to be doing 4X the resolution of DV then i want 4x the data. like say the nano, but i can barely afford the cameras/switching themselves. . . the price of doing full pro-sumer in HD is about double what DV was, and its compressed 4times more TOO.

loads of stuff on tv is still done in almost real time, they are not going to take 6 extra hours to fix everything afterwards in slow time, we TRY and do the same basic thing.

ONLY in the spirit of trying to get answers for the 35Mb/s stuff in these EX cams, i need answers that would be FOR this 35-25 rate, not for fully uncompressed things.

if the intent of the post was for FILM production, and not based on the EX cameras (be it film or not) then these answers should be in the cine area of the forum. and would probably apply across the board there no mater which camera is used.

a camera that has little or no compression doesnt apply to these suckers untill you add in something like the convergant design.

so IF the parameters were sucky 15&20-1 compression, then do ya do Kodachrome processing when the bits still exist or not?

also using a "way back machine" , "We'll fix it in post" was always a joke, 10 years ago before the great computer fixing was usefull and worked fair, it was highly recommended to not wait till post , and do it right the first time.

A few notes here:

I've managed to get 2 EX1s (and a Z1 or Z7 sometimes) matching up okay through careful setup and 'mutual white set' but there are differences between EXes. Maybe subtle ones, but it depends on how close you want the match without tweaking in post.

It's NOT like having a rack of CCUs and a vision engineer to ride the dials. Alas.

SD-SDI on EX1s? I have to say I wasn't overly impressed but it is what it is. It's not as good as a $1k HD-SDI to SD-SDI box from AJA or something. I am SO hoping that the EX1-R is going to do better.

You mention Theatrical lighting, and it sounds like you're no fan of the HVX 4:2:2 method. Well, XDCAM is by no means as bad as HDV with strong coloured lighting, but it still happens - easily calmed by nudging the chroma down later - but like you say, you don't want to spend time in post.

I looked long and hard at NewTek's SDI Tricaster which I think may have some paint controls at the head end, as well as some other pretty impressive tricks, but my market's not ready for it yet. Could fit yours?

Ok, when shooting 35Mbps on my EX1, I shoot with the color at about 70% saturation, and the luma reading about 15-90 IRE.

Yes, but not as much.

The word "right" is the tricky bit. What might be right for you, may not be right for someone else. And the choice I might make while shooting a 30 minute sit down interview might be quite different to me shooting a multi-day outdoor scene for a movie. If you can guarantee the weather, then maybe you can afford to shoot high-con, full sat. If the weather might change over the course of the day, or a couple days when you need to shoot, then the "right" way might be to shoot in a manner that will let you match the looks later in post.

oh great dont say that , wait what did you say :-)

i have Now becomes committed to the EX1r minus any return policy. using multiple EX1r things only, because many people already indicated that would be the only way to pass profiles so they all are set the same , so that i have covered (i think).

ummm, now i am in trouble, i thought that the HD-SDI outs on the EX cams was less compressed stuff? all good clean quality? was something not good about the EX1 SDI output?

but then i also forgot to ask anybody (oops) if the HD-SDI is post the internal camera processing. and now that you mention a switcher with color controls . . .

and with that "hi-sat" matrix on? , can a person just tweak up the color itself and leave the hi-sat Matrix and its lump sum change off?

HD-SDI out is very, very good. It passes muster for BBC, DiscoveryHD, NatGeo, etc. But it's not RAW. There is still some image processing taking place (like taking off that green hue you don't like so much), but it's minimal. And it's a far cry from what gets recorded on the SxS cards.

I've never used hi-sat.

Don't panic. The HD-SDI is very good, but the EX1's downconvert to Standard Def (I am NOT talking EX1-R, just the older EX1) is a little soft and lacking in oomph. So if you're committed to EX1-R, good on you. I pick mine up tomorrow, hopefully. And I will immediately get it home and check out its Standard Def credentials.

Yes, the profiles contain numeric data so two cameras have the same settings. However, and again I stress this is with EX1s, I found slight niggling differences between cameras when using pre-set white balances, but when using manually set white balances, it can take a couple of goes to get them the same on a monitor. But I, like many people, still cut in post using FCP multicam (or do it manually for 2-3 cams). So I can go the last mile in post.

Ho ho, it's not a CCU! Just some control over gain, Red and Blue in blacks and whites, and that was on the analogue model, not the SDI model which was unavailable for demo. But it is worth a look-see.

"I found slight niggling differences between cameras when using pre-set white balances, but when using manually set white balances, it can take a couple of goes to get them the same on a monitor. "

If that happened it would suck, in my mind , just from reading the manuel and here about the profiles and adjustements, i was planning on:
creating a SINGLE profile for everything (in my dreams)
using a preset white, shifting the WB slightly as needed
shifting the other Single color tint thing slightly as needed
then split screening and seeing 2 perfectally matched cameras.

then worring about the other 201 things that still need to be set up.

only because Manuel SETting of the WB is hard/impossible to do if the white card (huge) and lighting angles to both cams are not the same. AKA trying to get a white card balance has more possible chances for being different.

If you completely switch up profiles in the camera, are the changes to the processing of the color and contrast range and all reflected in the SDI output?