85 filter?

[Michael Brennan]

Graham has mentioned the camera is factory balanced to 5000k.

So will using a 85 filter on lens increase scope to grade images shot under tungsten light?

Or put another can you get away with not using 85 for run and gun jobs?


I'm still diapointed you didn't include an internal filter wheel with NDs and rotating polariser :(

Best of both worlds.



Mike Brennan

[Jim Arthurs]

I have had the same question for some time.

My current understanding is that, yes, you can get by without a CC filter in front of the camera and pull back a very good image.

Personally, I would shoot with CC whenever possible. I just think it will help keep away unwanted color tint on those blown out areas and improve general image soundness.

However, a new question based on yours, Mike... in REDCINE can you animate your grade parameters? Say you move from daylight to tungsten in the course of a take and choose to go without CC. Will REDCINE let you ramp from one set of correction to another?

[J. Bernard Vallon]

Everyone on the project describes the red as a big DSLR that can shoot 60 fps. In RAW recording, "their is no white balance", meaning you assign a white point in post. I've never had a problem doing that with RAW in any DSLR, I wouldn't image a difference in RED.

If the camera is "factory balanced at 5000k", Id imagine that means it is its "native" white balance. maybe an 85 filter would increase the quality of color slightly, but i dont know if it would be significant.

[J. Bernard Vallon]

Say you move from daylight to tungsten in the course of a take and choose to go without CC. Will REDCINE let you ramp from one set of correction to another?

that'd be neat

[Steve Gibby]

To me, the considerations are:

Shooting RC RAW
Image parameters are not baked in, and with the tweaking capabilities of REDCINE in pre-post, then filtering at acquisition is generally unnecessary.

Shooting RGB
Image parameters are baked in, and it's harder and time consuming to change them in post, thus filters during acquisition are a real consideration.

I've historically shot most of my footage "clean" because it simply gives me more potential uses in post, plus stock footage is a big part of my business plan - and clean footage has more flexibility and variety of licensing uses as stock. I'll shoot almost all my footage with RED One as RC RAW for that reason, and a few other reasons.

[Jim Arthurs]

Actually, to shoot "clean" footage with this sensor under tungsten WOULD be if using an 85 series filter... 85C would be the closest fit, methinks...

It would make a difference in critical areas, such as green screen work, etc... and those extreme manipulation moments...

[Evin Grant]

Were going to test this, but my feeling is that it will not make any difference
I might do some preliminary testing with the D2x and see what I can determine about that CMOS.

[Jim Arthurs]

Evin, it would be great to see a test on this... of course if your Nikon is not the same native C temp as the Mysterium, perhaps it is best to wait and test later.

Put me down in the "will make a difference" column if there are bets being laid...

It will show when colors start to saturate at different rates and in areas transitioning to pure blow out. I will have an 85C in the kit at all times for tungsten when using the RED.

Besides, if it is so easy to fix at lens with an 85C, why not do everything you can if you have the time?

[Evin Grant]

The only time I can see that making a difference is if there is some blow out/highlight cliping. But even then the highlight recovery tool in the RAW converter should do just as well. Although it's best not to clip anything if possible, and it does seem possible given the most recent tests. What should I be loking for as a difference Jim? When I convert the file I can choose any WB point I like so ideally I'll make both pictures neutral and then what, push the hell out of the saturation and see if one holds more info? It should at least give me a sense of what to look for right?

[Casey Green]

To me, the considerations are:

Shooting RC RAW
Image parameters are not baked in, and with the tweaking capabilities of REDCINE in pre-post, then filtering at acquisition is generally unnecessary.

Shooting RGB
Image parameters are baked in, and it's harder and time consuming to change them in post, thus filters during acquisition are a real consideration.



Speaking of RC RAW vs. RGB...

Jim, any chance the RGB feature will now be enabled for the first round release? We would certainly love to include that (as well as the RED B4 adaptor) in the tests at LART.

[Justin K Phillips]

If you're really worried about blowing highlights, underexpose. That's exactly what they did for the ISO 4000/8000 shots.

[Poi Boy]

In my extensive experience shooting raw digital images, I find that you only muck up the image if you use cc in front of the lens.
Aloha
-A

[Jim Arthurs]

Well, there you go. No CC and no penalties.

I still can't see how that's possible, but differ to everyone's greater RAW experience.

[Steve Gibby]

In my extensive experience shooting raw digital images, I find that you only muck up the image if you use cc in front of the lens.
Aloha
-A

I'll second that. I've used cmos sensor DSLRs extensively, shooting RAW, and I've never used a CC filter when shooting them - and the images and processing of them have been great - no problems. In technology, RED One most closely resembles cmos sensor DSLRs (with higher frame rates), so I'd anticipate the same approach to shooting with my RED One cameras. We'll see...

[Jim Arthurs]

Wait a minute, aren't we all bass ackwards here?

... the sensor is 5000K.

At the sensors native color balance, tungsten light would look overly orange/red to it, needing blue CCB in front of the lens to fix, not something in the warm85 series... we need 1800K of correction, but in the other direction...

?????

[Poi Boy]

You are essentially puting the filter on the image when you process the file.
-A

[Graeme Nattress]

Yes, currently balanced around 5000k. I'd not shoot with a filter though - I'd use the highlight tool in Redcine if you were to have a colour of light that caused a shift in the highlights (and as the colour gets corrected from un-clipped channels, you get some extra guestimated highlight detail too).

If you shoot with a filter, the kelvin white balance numbers won't make sense any more....

Graeme

[Casey Green]

I thought I read about this here before, but in theory, could you take a still or short "camera test" clip (RAW) while on set, off-load it into REDCINE make your adjustments, and then take that profile and load it back onto the Camera for use only on your Monitor output?

This would allow the DP and Director to have an idea what a specific adjustment would look like without needing any filters and also without needing to make any counter adjustments on the Assist Monitor at Video Village. (and yet still Record the RAW unfiltered clip). I suppose this depends on the limitations of what can be processed in real-time on the monitor output signal.

Is this possible? (or only possible for the actual Recorded Output)?

[Michael Brennan]

Wait a minute, aren't we all bass ackwards here?

... the sensor is 5000K.

At the sensors native color balance, tungsten light would look overly orange/red to it, needing blue CCB in front of the lens to fix, not something in the warm85 series... we need 1800K of correction, but in the other direction...

?????

Opps my mistake! should/could be 80A 80B or 80c depending on if you want 3200k or 3600k... not an everyday filter, which is why I asked the question, promted by Jims warning to Rookies to have ND filters.

Sorry for confusion..
And of course it is blue in colour.

As Gibby says recording RGB allows for less scope in grading and use of CC filter may be a good idea.
Some shooters have noticed that their is a little more scope in grading Viper filmstream material if on lens CC correction is used to remove the green bias.
Green channel reaches saturation point ahead of the other channels without the cc filter.

But then RED is more of an anaconda than a viper so it may deal with 2500cc shift in its stride.


Mike Brennan

[Jannard]

We think you will NOT want to add filters when you shoot the RED ONE in tungsten. 1st off, if you did add a filter, it should be an 80B (or thereabouts)... not an 85. But the light loss is worse than the correction. We are going to do complete testing on this in the next few weeks.

Jim

[Jim Arthurs]

Damn good to know the answer to this. Yeah, blue filtation sucks light up like a big dog. My entire thinking path on this thread was with the much weaker 85 series going the other direction.

I love being wrong when what I'm wrong about benefits me. :)

[Matthew Rogers]

We think you will NOT want to add filters when you shoot the RED ONE in tungsten. 1st off, if you did add a filter, it should be an 80B (or thereabouts)... not an 85. But the light loss is worse than the correction. We are going to do complete testing on this in the next few weeks.

Maybe I missed something Jim, but how are we going to see the "correct image" on a set? Am I correct that we are going to be able to "white balance" and that setting will color correct to the monitor and also put the number in the meta data for later CC?

Matthew

[Steve Gibby]

Maybe I missed something Jim, but how are we going to see the "correct image" on a set? Am I correct that we are going to be able to "white balance" and that setting will color correct to the monitor and also put the number in the meta data for later CC?

Matthew

"The one operational clarification I'd offer to this is that the images through the viewfinder and monitor ports can look off color if you don't set white balance or hit the Auto White Balance from time to time. Again this does NOT affect the RAW data, and WB information IS stored with the RAW data as metadata, but adjusting (or having the camera recalculate) white balance as you shift between different color temperature lighting conditions - such as indoor to outdoor - is the only way to prevent a color cast in your viewfinder and on your on set monitors."

Stuart English
5/3/07
RED User
http://www.reduser.net/forum/showthread.php?t=2196&page=3

[Graeme Nattress]

Yes, through the RGB pathway, monitoring appears correct (and if I do say so myself) and rather pretty.

Graeme

[I Bloom]

Here is my understanding about color images and RAW Bayer Pattern, which comes not from RED but from my DSLR.

It's all about the green channel. The ideal RAW Image has the most information or the best histogram on the green channel. Why? It's really about demosaicing.

It starts like this: Most images found in nature have more variance in luminance than in hue. Hue is determined by the ratio of color values at a given pixel value. Using the bayer filter we do not know for example the specific Blue value on a pixel that is filtered Green but we have a very good guess of that value because we know that the ratio of Blue/Green values between two neighboring pixels is going to be very similar. So the Green value will help us guess the blue value. A 4K Bayer Pattern chip camera is better than a 1.5K 3 Chip camera (without pixel shifting) because the overall luminance values across the image can be reconstructed with decent accuracy.

There are many different approaches to demosaicing a Bayer pattern image. We use complicated algorithms to try and fill in the blanks. But because there are twice as many green pixels in a checkerboard pattern most algorithms start by interpolating the green channel and then use it as a guide for the red and blue channels.

The practical upshot of this is that heavy color filtration that effects the green channel will effect the demosaicing process, thus effecting the percieved detail quality of the final image. An image that is lacking variance (a full range of exposure values in the green channel) might suffer from issues that have little to do with color. A constricted or noisy green channel might interpolate the entire image poorly.

So this is really going to take some testing to know this for sure but my intuition is that in a situation where the RGB Parade shows green is strongly shifted in one or another direction it makes sense to correct color in front of the lense. Even if you will shift it back in post it may be better to do this work after interpolation.

That's my feeling. I'd love to see Gibby's team do some tests that address this specific issue.

IBloom

[Graeme Nattress]

Yes, green is important, but algorithms can be made so as to be very tolerant of the image data and do a good job pretty much no matter what.

Graeme

[I Bloom]

Yes, green is important, but algorithms can be made so as to be very tolerant of the image data and do a good job pretty much no matter what.

Graeme

Can you say off hand which algorithms will be implemented in REDCine?

IBloom

[Paul Leeming]

Hopefully this quesrtion/suggestion is not taking this thread off topic, but with RAW colour correction is there some function in RedCine to take an image of a GretagMacbeth ColorChecker chart and quickly balance the image to it? I've seen an addon to Photoshop that does this but it costs money and you'd think Photoshop would have this built in:

http://www.pictocolor.com/incamera.htm

This would be a great addition to RedCine if it could be implemented!

Cheers Red Team!

Paul

[Graeme Nattress]

The colours are defined to the camera already. All you have to do is pick a white point.

Algorithms - I've written about 20 so far. We're using "nice" as our goal - overall image, rather than going for extreme sharpness. If you aim for extreme sharpness you get "bitty" results that just don't look good under motion. If you liked what you saw at IBC on the big screen, you'll like what we're doing.

Graeme

[Michael Brennan]

The colours are defined to the camera already. All you have to do is pick a white point.

Algorithms - I've written about 20 so far. We're using "nice" as our goal - overall image, rather than going for extreme sharpness. If you aim for extreme sharpness you get "bitty" results that just don't look good under motion. If you liked what you saw at IBC on the big screen, you'll like what we're doing.

Graeme


Graeme is it feasible and desirable to have the algorithims user defined (or custom made at the factory) for specific shooting environments where the camera isn't doing double duty.

ie
Bluescreen
Underwater
Aerial

For aerials for instance we would be interested (most of the time) at maximising resolution of fine detail (often foliage) and not so interested in sensitivity

Perhaps Underwater we would wish for sucking in as much red/yellow as possible.

For Bluescreen perhaps greater tonality of blue (if you get my drift) to allow easier removal of spill.


Does tweaking algorithums do much to the above list?



Mike Brennan

[Graeme Nattress]

I don't think demosaic effects much of the above, other than resolution for your landscape. Problem there is put too much detail in, you might get buzzing or artifacts as the detail moves, which is not to bad for a DSLR still where you can go in and hand tweak, but that's not suitable for movies - we've got to look at the time dimension too.

Graeme

[I Bloom]

Graeme is it feasible and desirable to have the algorithims user defined (or custom made at the factory) for specific shooting environments where the camera isn't doing double duty.


I think all Debayering algorithms are implemented in post, via REDCine. So you'll have your choice and don't need to worry while you're shooting. As far as I can tell the camera itself only downscales images from the bayer pattern, a much more straightforward task. NOT debayering in camera, while still compressing, is part of what makes RED's pipeline so special.

IBloom

[Graeme Nattress]

We do full debayer and downscale in camera for the RGB recording pathway for best quality. For RAW recording, it's done in post.

Graeme

[I Bloom]

We do full debayer and downscale in camera for the RGB recording pathway for best quality. For RAW recording, it's done in post.

Graeme

Do you debayer to 4K and then downscale?

[Graeme Nattress]

In camera, yes.

Graeme