Welcome to our community

Be a part of something great, join today!

  • 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 :)

What did I do wrong here?

Rakesh Malik

Well-known member
Joined
Apr 8, 2013
Messages
1,029
Reaction score
1
Points
38
Location
Vancouver, BC
Here are three frames from a music video shoot I did last week. I was filming with an Epic-W, using Zeiss lenses. Due to having not much light for this part, I raised the ISO to 3200 (making up for not having as much light as I'd have liked, due mostly to budget constraints, and such).

I was also stopping down a bit to gain some depth of field to make the artists easier to follow with focus.

In one frame there, the smoke is nice and clean looking, when the light on it is white. In the purple one, the smoke looks very blotchy, and in the third frame there's quite a bit of shadow noise in the frame.

My first thought about the blotchy purple is out of gamut color; I'm not sure that I'm correct about that. I had the color set to DragoncColor2 and RedLogFilm.

I recorded these in ProRes 422HQ, and extracted these frames without grading. They're recorded in 4K, but we're planning on mastering in HD, so it's not going to cause it any real harm on this shoot, so I'm really just looking to learn from whatever mistakes I made... once i find out what they are :)

So I learned after that shoot (naturally) that DragonColor is a better choice than DragonColor2.

Is this possibly a side effect of not rembembering to black shade or something?

Thanks!
 

Attachments

  • purple.jpg
    purple.jpg
    93.3 KB · Views: 0
  • white.jpg
    white.jpg
    91.6 KB · Views: 0
  • shadow.jpg
    shadow.jpg
    89 KB · Views: 0
First frame looks out of Gamut for sure, possible clipping the blue channel? -

The one with the guy - (i'm assuming you mean his face & arm) just looks underexposed on a sensor level. - But if you grade it darker it will probably mask most of the noise. - 1080 will help as well. :)
 
If the shadow noise is just underexposure, then that's easy enough to deal with as long as I have enough light. I was just wondering if maybe I should have remembered to black shade or something like that.

Any advice on preventing the out of gamut part when using ProRes?
 
I would look at the material on a vectorscope within (say) Resolve and make a decision on how far the colors can be pushed before they're illegal.

Note that RedLogFilm is not ideal for viewing, and you may get a false sense of what's in the picture. I would look at one of the RedGammas or DragonGammas and see how it looks with a calibrated Rec709 display.
 
The blotchy purple is DragonColor2, otherwise it seems underexposed and recording in 422HQ seriously cripple your grading options. If you are shooting ProRes for final output, always go for 4444. Since Epic-W can't do 4444 in 4K, you should shoot full 8K or 8K "HD" (if you want to max fps). The whole purpose of Red is the R3D raw format, it's not really meant to be a ProRes camera like other brands. This is why ProRes recording can lack in quality in comparison. Only time I shoot ProRes is during long takes like interviews etc.
 
i think there is an anti raw mindset from the arri raw experiences people have had? red raw is significantly easier to work with
 
Recording prores only on helium is a bit of a risky move as it does not allow the wide gamut RGB stuff. Simply you got hit by out of gamut colors and not much you can do about it at this point than trying to grade it to look better if you don't have the r3d´s. I think when RWGRGB is available in camera then recording prores will be a very good option especially when using the full sensor and especially when shooting in the dark as it´s simply downscales before compression and the RWGRGB is a very nice base to work from. But until we got that in camera I recommend to record also the r3d.
 
Yeah this seems like one of the big downsides to incomplete color science on Helium, really cripples prores since color errors can only be fixed with a special raw workflow using RWGRGB...
 
Yeah this seems like one of the big downsides to incomplete color science on Helium, really cripples prores since color errors can only be fixed with a special raw workflow using RWGRGB...

Yeah is definitely a bummer. Though *Brent did [sneakily] "confirm" this month we'll be seeing the new color science.

EDIT: My hopefulness got the best of me -- Brent confirmed "More exciting stuff this month though" in response to new CS talk here.
 
Last edited:
I have to say your biggest mistake is the under exposure. You said you had to raise the ISO already as you didn't have enough lights. To stop the lens down for focus is a fatal error and a complete technical misuse of the equipment. What follows from that can only be bad results...
 
I have to say your biggest mistake is the under exposure. You said you had to raise the ISO already as you didn't have enough lights. To stop the lens down for focus is a fatal error and a complete technical misuse of the equipment. What follows from that can only be bad results...

Seems pretty harsh to me. To stop the lens down for DOF is something used by many, many cinematographers. Sounds to me like this is one of those projects where having more lighting and/or AC with WirelessFF was out of the budget. That just happens sometimes.
 
I have to say your biggest mistake is the under exposure. You said you had to raise the ISO already as you didn't have enough lights. To stop the lens down for focus is a fatal error and a complete technical misuse of the equipment. What follows from that can only be bad results...

That's a pretty misguided and arrogant statement, given that you weren't there...

At least everyone else here has been informative.
 
Recording prores only on helium is a bit of a risky move as it does not allow the wide gamut RGB stuff.

On this shoot, the main reason for recording to ProRes and not R3D was for recording time. It was a last minute thing and honestly I didn't even know about it until the day before; I had no idea what we were going to be shooting... and I was also lead to believe that we were going to be outdoors during the day, rather than indoors at night shooting with colored LEDs backlighting smoke.

This footage looks good enough for HD masters, which is the target. I do want to be able to deliver ProRes or DNxHD down the road though; it's a useful feature that sounds like it's going to become a whole lot more useful when the new image processing pipeline is available.
 
Never shoot ProRes on Epic-W until the next firmware. You are essentially baking in a missed debayered color shift, you need to shoot raw. Also never use Dragoncolor2 when there are any spikes of blue/red's (aka purple) as it accentuates those colors harshly and doesn't process properly on Helium. If this was raw it would be an easy fix by changing that to Dragoncolor or RGBwidegamut, but since you shot ProRes I don't think you can fix that shot unfortunately.

As for noise you are shooting underexposed at 3200 ISO, Helium is not an A7S... it's not quite clean at 3200. But that's an easy fix with neatvideo
 
That's a pretty misguided and arrogant statement, given that you weren't there...

At least everyone else here has been informative.

Rakesh,
While Daniel's delivery was a bit harsh, the info he was conveying is correct. The concept of ISO on RED is somewhat misleading coming from stills (which I did a few years ago and enjoyed every mistake along the way). Forgive me for assuming you aren't 100% familiar with how RED ISO works, but it isn't the same as a DSLR or many other cameras which merely boost the gain on the sensor. Practically all of my early Scarlet-D footage was grossly underexposed for this exact reason, because I was applying DSLR logic to a RED. The RED approach lies in redistribution of middle gray within the 16ish stops of sensitivity. It is quite confusing until you've trained your mind to think of it in this way. Increase ISO, but never stop down afterward. Exposing heavily to the right will give you a fatter negative to work with. Your highlights will hold better than you think. In fact, there's an approach to using ISO in low-light situations in which you dial the ISO DOWN to assign more stops of sensitivity to the lower end of the tonal range and then open up the image in post (you can apply a LUT in the display so you can see the image). Conversely, you can increase ISO in bright scenes with delicate highlight tones (think clouds or wedding dresses outside) to assign more stops of sensitivity to the higher end of the tonal range. I've not tested the former of these two approached, but in theory it makes sense – give more sensitivity to the part that needs it.

Pay close attention to your exposure tools – GIO, false color, histogram, goal posts, stoplights – and it's hard to go wrong.

There are dozens of ISO threads, most of which I've sifted through to get a better understanding of how this system works and at times it's still baffling.
 
That's a pretty misguided and arrogant statement, given that you weren't there...

At least everyone else here has been informative.

I am not sure I would call Daniel misguided (I don't know if he is arrogant). You should take a look at his body of work, he is pretty good at this shit. Ridley Scott likes him anyway I guess.
 
Rakesh,
While Daniel's delivery was a bit harsh, the info he was conveying is correct.

Not entirely. Stopping down was the right thing to do under the circumstances, because the artist was a complete newbie and his movements were very random. The only reason I raised the ISO instead of adding more light is that I didn't have access to more light to add.

The concept of ISO on RED is somewhat misleading coming from stills (which I did a few years ago and enjoyed every mistake along the way). Forgive me for assuming you aren't 100% familiar with how RED ISO works, but it isn't the same as a DSLR or many other cameras which merely boost the gain on the sensor. Practically all of my early Scarlet-D footage was grossly underexposed for this exact reason, because I was applying DSLR logic to a RED. The RED approach lies in redistribution of middle gray within the 16ish stops of sensitivity. It is quite confusing until you've trained your mind to think of it in this way.

It is something I'm still getting used to.

Increase ISO, but never stop down afterward.

I stopped down to increase depth of field.

Exposing heavily to the right will give you a fatter negative to work with. Your highlights will hold better than you think. In fact, there's an approach to using ISO in low-light situations in which you dial the ISO DOWN to assign more stops of sensitivity to the lower end of the tonal range and then open up the image in post (you can apply a LUT in the display so you can see the image). Conversely, you can increase ISO in bright scenes with delicate highlight tones (think clouds or wedding dresses outside) to assign more stops of sensitivity to the higher end of the tonal range. I've not tested the former of these two approached, but in theory it makes sense – give more sensitivity to the part that needs it.

I understand the basics of how ISO works on Reds, but I'm just getting started. And the legion of exposure tools available is taking some getting used to; there are a lot of them. Red ISO does seem pretty counter intuitive. I somewhat understand it, but I haven't used the camera enough to be able to previsualize what the settings will do yet.

Pay close attention to your exposure tools – GIO, false color, histogram, goal posts, stoplights – and it's hard to go wrong.

There are dozens of ISO threads, most of which I've sifted through to get a better understanding of how this system works and at times it's still baffling.

Yeah, I do forget about some of them now and then, since they don't exist on any of the cameras I've used in the past.
 
Never shoot ProRes on Epic-W until the next firmware. You are essentially baking in a missed debayered color shift, you need to shoot raw. Also never use Dragoncolor2 when there are any spikes of blue/red's (aka purple) as it accentuates those colors harshly and doesn't process properly on Helium. If this was raw it would be an easy fix by changing that to Dragoncolor or RGBwidegamut, but since you shot ProRes I don't think you can fix that shot unfortunately.

As for noise you are shooting underexposed at 3200 ISO, Helium is not an A7S... it's not quite clean at 3200. But that's an easy fix with neatvideo

So does downscaling it to HD like we'd planned for this shoot anyway, so on this one we're fine. And now I know better. :)
 
Stopping down was the right thing to do under the circumstances, because the artist was a complete newbie and his movements were very random. The only reason I raised the ISO instead of adding more light is that I didn't have access to more light to add.

I stopped down to increase depth of field.

The bottom line here is that raising ISO isn't the solution for not having enough light, as it is with a DSLR or most any other video/cinema camera.

Stopping down to increase depth of field after increasing ISO will always leave you with an underexposed image, which is absolutely counterintuitive. More depth of focus was the correct thing to do intuitively given what you were shooting, but ultimately left you with a light-starved sensor. You need to add light, which in this instance wasn't an option. Just about the only thing you could have done is increase your shutter angle to let more light in which would have introduced more motion blur into your image.

As someone else mentioned, Neat Video may be your friend in this situation.
 
The bottom line here is that raising ISO isn't the solution for not having enough light, as it is with a DSLR or most any other video/cinema camera.

Stopping down to increase depth of field after increasing ISO will always leave you with an underexposed image, which is absolutely counterintuitive. More depth of focus was the correct thing to do intuitively given what you were shooting, but ultimately left you with a light-starved sensor. You need to add light, which in this instance wasn't an option. Just about the only thing you could have done is increase your shutter angle to let more light in which would have introduced more motion blur into your image.

As someone else mentioned, Neat Video may be your friend in this situation.

I should have tried increasing the shutter angle.

ISO on Reds is pretty counterintuitive. But I'm learning. That was the point in joining that shoot; the budget didn't really justify an Epic-W, but I figured it was a good opportunity to get some camera time in with it just to get more familiar with it and its options.
 
Back
Top