Redcine blacks coming in at 10 IRE in FCP

[Philip Powell]

Hey Guys,

We've been an Avid Symphony based house for a long time and only recently switched over to FCP. So I fully admit I'm floundering a bit.

In any case, we've been pulling in the proxys in FCP and they are fine, that's to say, I've got blacks at zero and highlights near 100.

But when I take the same R3D files, process them in Redcine on a PC, to 1920 1080, Camera RGB, Rec 709 using a varity of codecs, AJA, Animaton etc. Full quality.

They come in with their blacks at 10 IRE (it doesn't appear to be clipping the blacks, although my eyes could be getting fooled) It's like all of the info was pushed up, from zero to ten.

Could this be something I've got set wrong on the import side from FCP?

 

[Graeme Nattress]

The raw data fully preserves the entire range of data captured by the sensor, hence you can see all the data, slightly raised.

Graeme

 

[timecodemultimedia]

Graeme is correct.

You get all the data and all the latitude to push or crunch the signal at your disposal in color correction. Generally, I much rather work in a flat environment that doing one-light cc out in RedCine (for example) which is limited in comparison a color correction system.

I have much better controls of color correction after the footage has been edited or converted.

 

[Antoine Fabi]

Oh...

I thought that black calibration was making abscence of light = 0 value equivalent...

...anyway, it works perfectly.

 

[Philip Powell]

OK, trying a few more tests to understand things better

OK, trying a few more tests to understand things better

Using build 15 I shot a short clip with the lens cap on, at 4K 2:1. When I pull the proxy into FCP, it comes in with blacks mapped to zero IRE. (as you would expect)

Same clip, when I import it using log and transfer into FCP, it comes in just the same, with black at zero IRE.


But if I take the clip into Redcine and (using Camera RGB, Rec 709) and output it, using various codecs, the blacks come in at 10 IRE. The historgram in Redcine seems to reflect this, as most of the blacks are just above the last zone.

When I test regular images with a full range that I've shot outside I get similar results, blacks boosted, highlights boosted as well but to a lesser extent.

I guess I'm confused, as I would think (although I know the proxy is just a quick debayer) that the same file processed through Redcine for Rec 709 (without any curve tweaks) would look very close to the proxy, but it doesn't.

Does this mean to get zero IRE black all shots coming out of Redcine must be adjusted somewhat if they headed for Rec 709? Not a big deal, just trying to understand.


Can someone enlighten me as what is occuring? Thanks.

 

[vsrecord]

I'm brand new to the red workflow and I'm see this exact same thing happening with my QuickTime exports from Redcine. The proxies look fine and have a nice range in the waveform monitor. In Red cine the footage looks fine, when I export a quicktime in ProRes format, that's all I've tried, the footage comes in with the blacks raised. So is this normal and to be expected? My client thinks I'm doing something incorrectly, but I'm doing exactly what's described above.
Thanks for any enlightenment.

 

[Philip Powell]

The treatment of proxies vs. Redcine output

The treatment of proxies vs. Redcine output

The raw data fully preserves the entire range of data captured by the sensor, hence you can see all the data, slightly raised.

Graeme

What I'm trying to figure out though is what is happening differently between import of the proxies vs. straight untouched output of Redcine in the same color space, referencing the same R3D files.

Again, taking my clip, that I shot with the lens cap on. (total black)

The proxies come into FCP with black at zero IRE.

The same file processed through Redcine comes in with black at ten-ish.

They're both in the same color space, referencing the same R3D files, it would seem to me that they would come in mapped similarly, either both at zero, or both ten.

I know that there is more range recorded in the R3D file, and that I see inside Redcine, in both the blacks and the highlights, in it's 0-1023 range.


I'm trying to speculate as to the cause of this difference.

Is it that there is minute information recoreded below black (theortically zero) but Redcine uses, and maps to zero, hence my raised 10 IRE blacks?

If true, again, not a big deal, but it seems to me then that most of the shots that passe through Redcine (intended for Red 709) would have to have it's blacks adjusted, as Redcine is pushing them/mapping them slightly higher than they should be.

Any insights would be appreciated. Thanks.

 

[vsrecord]

I did a little more testing today and came up with a few things:
1. The ProRes that comes back has all the values shifted upwards, enmass.
2. If I park on identical frames in the original sequence and the ProRes sequence and switch back and forth with the WVM open can see this shift. Also the color target lines in the WFM and the color targets in the Vector Scope shift.
Which seems to indicate that the color space for the two sequences is different and Final Cut is not changing its output, that would account for the shift.
Using A plugin called "The Grading Sweet", which is like the three way CC but instead of three color wheels it has Red, Green, Blue adjustment sliders.
If I pull down the mid and then stretch the high and shadows I can get get the WFM to look almost perfect. Then looking at the Vectorscope I adjust the color sliders, just a pinch of red and remove a touch of blue and the prores and R3d file match when I switch. From there I continue on with my color correction.
Is this the correct way to do things? I doubt it, but I like the look of the footage as displays from the proxy sequence so I figure its a good place to start color correcting and the final output looks great.
If I were doing a higher end project I'd like to know that this is truly what I have to deal with.
I'd really feel more comfortable if I knew that Sir Pancake and weren't the only ones seeing this.

 

[Philip Powell]

I did a little more testing today and came up with a few things:
1. The ProRes that comes back has all the values shifted upwards, enmass.
2. If I park on identical frames in the original sequence and the ProRes sequence and switch back and forth with the WVM open can see this shift. Also the color target lines in the WFM and the color targets in the Vector Scope shift.
Which seems to indicate that the color space for the two sequences is different and Final Cut is not changing its output, that would account for the shift.
Using A plugin called "The Grading Sweet", which is like the three way CC but instead of three color wheels it has Red, Green, Blue adjustment sliders.
If I pull down the mid and then stretch the high and shadows I can get get the WFM to look almost perfect. Then looking at the Vectorscope I adjust the color sliders, just a pinch of red and remove a touch of blue and the prores and R3d file match when I switch. From there I continue on with my color correction.
Is this the correct way to do things? I doubt it, but I like the look of the footage as displays from the proxy sequence so I figure its a good place to start color correcting and the final output looks great.
If I were doing a higher end project I'd like to know that this is truly what I have to deal with.
I'd really feel more comfortable if I knew that Sir Pancake and weren't the only ones seeing this.

I guess you and I can keep talkinig about this till we figure it out.

I have noticed something slightly different than you. I don't notice the IRE shift enmasse, it's different on the low and high end. While it's true that the whole scale shifts, the blacks move MORE than the highlights.

Black moves like 8-10 IRE up, but the highlights move less, more like 3-4 IRE.
I suppose really this question should be asked in the Redcine section as this issue has more to do Redcine than FC.

I'll keep testing and see if I can figure anything more out.

I'm not saying Redcine is doing anything wrong, just trying to understand WHAT it's doing so I can make reasonable decisions.

Anyone else every done a black test like this, through Redcine?

Thanks.

 

[vsrecord]

I think I'm getting the hang of this now, its the proxies that are not displaying correctly. I checked the historgram of a shot in Redcine and it matched the pro res and uncompresed outputs historgram. When I changed the gamma curve on the shot to match the proxie curve in FCS things the pro res shot matched the proxy.
So after editing my proxies and sending them back to Redcine for transcoding to prores, if I set my curves to where I like them my shots are looking much better in FCP.
So essentially if your transcoding to Quicktime a one light is a necessity.
Hope this helps.

 

[Philip Powell]

I think I'm getting the hang of this now, its the proxies that are not displaying correctly. I checked the historgram of a shot in Redcine and it matched the pro res and uncompresed outputs historgram. When I changed the gamma curve on the shot to match the proxie curve in FCS things the pro res shot matched the proxy.
So after editing my proxies and sending them back to Redcine for transcoding to prores, if I set my curves to where I like them my shots are looking much better in FCP.
So essentially if your transcoding to Quicktime a one light is a necessity.
Hope this helps.

Hey,

Yes, I don't think it's that the proxies are coming into FCP correctly, or incorrectly as it were. But if you use your IRE meter while shooting with the lens cap on, it will read O IRE on the camera, and if you bring that same clip into FCP directly via proxy or log and transfer it also reads as O IRE.

But that same clip pulled into redcine, the blacks are at about 10 IRE, and will reflect that no matter the codec used to output to FCP.

To me, I'm just trying to figure out a proper workflow, but as you've stated, it seems to me that viturally any shot pulled into Redcine, would require a curves adjustment to get the blacks a bit deeper. (assuming you're going to Rec 709)