Welcome to our community

Be a part of something great, join today!

Log in
  • 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 :)

Grading RED footage...

studying noise

studying noise

stevesherrick said:
There seems to be some compression blocking that is evident in the blacks. But this looks like JPEG compression, so I'm just going to ignore it unless there was a TIFF to compare to.
Click to expand...

Anywhere specifically you see this? I've looked around this image at 400%, using extreme level shifts to study the dark areas, even doing strong sharpening to bring out any non-uniformity. When you bring up the levels, the darker tones have more noise than the brighter areas (of course) but I can't find any artifacts. The noise looks like smoothly textured, random noise to me.
 
Derth,

I have been doing some of this experimenting as well. What I have been trying to do (instead of attempting to create looks in post) is to create looks or guidelines for different shooting situations.

IE:

STRAIGHT TO FCP and out the KONA to TAPE:

If I know I am going to go straight to FCP with the proxies or rendered media, then I try to get the best look I can in the camera (one that will give me some good room for CC in FCP but doesn't force me to go back to the r3d files and do the DPX thing.

FILMOUT:

Here you want to take your tests right through a film workflow. Exposing the sensor according to how you might get the best Dynamic range and latitude for going to a FILM recorder.

BEST OF BOTH

I am also testing for a look that will give me good wide histograms in the QT files while protecting the highlights (and information) so if a film out is needed down the road, you've got it good both ways.

I think that this approach is good because the camera can do so many things (and we need it to do so many things) that it just becomes important to run tests and see how the sensor responds to different situations.

We've tested, tungsten, daylight, various ISO, interior, exterior, etc. Some jobs (the ones you know are going straight to video, you might treat a bit differently.

Overall, I think that this sensor likes a lot of light (or signal) and I tend NOT to underexpose. I go after a good solid histogram, and then god things flow from there.

I am not sure about the descrepencies between RED ALERT and LOG and Transfer but I suspect it has something to do with REDCODE. Have you noticed that the files that you put into an FCP timeline are shift from light to dark when you hit play and pause.

My guess is that the first version of the LOG and TRANSFER tool is going to yeild the best "quick and dirty" results until we bring these sorts of problems to then attention of the RED team so they can do some pixel voodoo.

Right now it is "working" but it has a way to go before it is really elegant and I can see how much potential it has right now.

I'd love to see a log and transfer tool morphed with RED ALERT that would allow you to do a small correction pass on the initial transfer with a choice of FULL or HALF debayer quality right at that stage. Then I could go for FULL quality and a one light in one piece of software have my material ready to edit and do a straight to tape finish. That would be elegant for a lot of different situations.

I'll do a couple of tests and report back.

David
 
Link to 4K backlight woman frame?

Link to 4K backlight woman frame?

Jannard said:
The examples below are not designed to be pretty... it was to show RED ONE footage in the most difficult situation. Make your subject the darkest object in an extremely high range scene... and hold tone everywhere.
Click to expand...

I did not see the link to a 4K TIFF of the frame used in this example, could you please make a link for a 4K TIFF of this frame so that I can try some things on it and later try it as an example hard to print image in a film recorder?

Thanks in advance for your help.
 
www.DANCAD3D.com said:
I did not see the link to a 4K TIFF of the frame used in this example, could you please make a link for a 4K TIFF of this frame so that I can try some things on it and later try it as an example hard to print image in a film recorder?

Thanks in advance for your help.
Click to expand...

We aren't going to post any tiffs from Build 16 until we are done.

Jim
 
I've been playing around with the curve solutions posted here and have come to the same conclusions I assume everyone who owns a RED and has done some testing has come to: if you want your highlights to blow out like video, they can - and, if you want your highlights to roll off like film, they can.

Did some tests exposing at 640 like Jim recommended (effectively underexposing my image), and here is what I got:
aaron1.jpg

Highlights are overexposed and burn out as they do (ala video).

But... open up RedAlert - move ISO to 250 and apply an S curve similar to the one Evin posted above:
aaron2.jpg

You quickly discover that the highlights weren't ever actually blown, but bumping the ISO up in camera takes ALL of the levels up giving you that dirty video BURN (which is what REDspace addresses and what MacGregor was talking about a while back).

This test may be redundant (as Evin posted a great example above), but I just wanted to second that you can roll off whatever way you please if you know how, and - if build 16 helps make this easier: great news.
 
Completely confused thread--

It keeps flipping between the ISO is only metadata, and the ISO is like a "gain" control.

It can't be both.

WHEN SHOOTING

Does changing the ISO setting in any way change the data coming into the RAW sensor?

Is the ISO setting in fact a "gain" control?

MONITORING

Or, is the ISO setting strictly changing the look of the viewfinder so that the user will adjust shutter and iris depending upon what he sees on it?

From what I have read, it seems to be the latter, and there seems to be no "gain" control affecting the recorded image in any way. Therefore the concept of "gain" should be tossed out as misleading and confusing. Is this correct?
 
It is only metadata. This is not in question. There is NO gain control whatsoever.

It only affects how you monitor.

The most important element is to expose properly. This mean iris, ND and shutter. Choose your ISO on set, pick the right lens and the right amount of light to feed the sensor, and off you go.

The above example shows that setting the right ISO in RedCine/Alert is extremely important. This is not the same as setting it in camera.
 
Hello all!

Very new to posting here since I mostly read and take in information ;)
Im working with the RED ONE camera only in my business but yet I have used the camera myself little. Its a pity but I deal with footage from RED every day and I notice the difference between film and digital that can be reduced very much. Many people say that digital is meant to have a different look and feel than film but RED is here to give the same feel, look and vivid image as film will give without the hassle. I love the look of film and I think you can achieve a very lively image with RED as you do with film but with the expense of time and experience (money too if you are a colorist ;).

Just by using Photoshop I graded these:
4kdrinkly8.jpg

conferencecmixerfa2.jpg

strandwg6.jpg


Images watched on a PC or a MAC computer will differ in light I suppose but keep in mind mind this is just experiments. I work a lot with the roll-off that is a bit different from film and RED.
Though just to know how the RED camera captures images and light you get an advantage in grading. Since I have not made a perfect system to conform it to "filmlike" image before grading I don't want to reveal my tricks yet but I will show you soon if anyone is interested. I want to thank the RED team for bringing the light to the artists around the world! Very nice to hear from you mr. Jannard here personally on the forum. RED is like mind to matter (very psychadelic description but I hope you will understand me).

Thank you
/Hed

EDIT: added one more image, graded the last one a bit warmer.
 
Jonas Hed said:
Hello all!

Though just to know how the RED camera captures images and light you get an advantage in grading. Since I have not made a perfect system to conform it to "filmlike" image before grading I don't want to reveal my tricks yet but I will show you soon if anyone is interested. I want to thank the RED team for bringing the light to the artists around the world! Very nice to hear from you mr. Jannard here personally on the forum. RED is like mind to matter (very psychadelic description but I hope you will understand me).

Thank you
/Hed
Click to expand...

Hi Jonas, please reveal your tricks. I and I am sure many others will be interested. Thanks

Lim
 
Jonas, YIKES, not to sound like I'm trying to hurt or feelings or anything but... I mean this in the nicest way possible but your saturation is neon. I know color theory and correction is intensley based on objective decisions and tons of real world experience but...seriously man...The first image (milk girls) has been already available in a corrected or nearly corrected state since they put them online. In that image and the third image from Mike Curtis' website (or at least he hosted that series of pictures from spain) you've not only over over saturated but blown out the upper range of the pictures. I'm trying to grasp at what you point was but I'm having a hard time. Please explain. :)
 
Well Jenga in my post I said that these were experiments and if you are watching it on different computers you will see different neuances and contrast. I have also been a bit experimental and rough with these and maybe should have pulled the saturation down more.
The first image (milk girls) has been already available in a corrected or nearly corrected state since they put them online.
Click to expand...
Maybe it is a light grade on it but what is your point? Could you not CC a graded image? Is there a certain rule? Please be more specific.

The "Spain" image is not blown out but sure it is light in the upper part.
I think you have to do something about the settings on your display if it gets neon. I can agree is over saturated (in this "experiment" ;) ) but nothing is "neon" on my display. Im on a pc and I have not watched them on a mac but in my experience the mac will show a lighter image than on PC if its not calibrated or gamma corrected. Please do not say it is blown-out if it is not. There is still info in the WHOLE image.
I'm trying to grasp at what you point was but I'm having a hard time. Please explain.
Click to expand...
If you read my post you will notice that my point is: you can do many different looks with the RED ONE image and get huge amount of info to work with from it. Think my images just proved some of that by making you believe they were blown-out ;) (joking)

Im saying this in a nice way too so do not take it the wrong way :turned: .
If you want to give me some hints or tricks of your CC experiences to anyone you are much welcome.

Thanks
/Hed
 
Of course. I was just confused because it seems no matter what display I look at these on, they are beyond what i would consider a normal range of experimentation in the coloring dept. (I've checked it on a calibrated Mac Pro, a default settings PC and even a Wii hooked up to a sony LCD tv) but again, it's all about personal preference and I would certainly assume you are experimenting before i would claim that your own color profile is off or something like that.

Good luck in your cc explorations :)
 
battistella said:
It is for this reason I have been asking if a program like Red Alert can have a V-out. Programs like FCP, AE, photoshop, motion, color, or many other third party programs provide a way to be displayed on an output monitor . . .

David
Click to expand...

1) Monitor out in RedCine, etc., would be great. Also, when relaunching the app, can we have it bring up a project dialog box instead of the current default of loading the last project?

2) Regarding saved color settings in RedCine . . . when a color preset is loaded, can the name be displayed on screen as the curently applied preset?

3) In file management in RedCine, can the last directory accessed become the default when loading footage into the library or loading / saving a color settings preset?

Thanks! Great job Jim and team!
 
blue channel noise

blue channel noise

"The sensor is Daylight balanced. If I shoot tungsten, I still work with a daylight setting in RED Alert! and correct later. Low light will also not yield as good of range."

Jim,

Is it possible to reduce noise by using a more blue balanced bulb greater than 5500K?

Are there tests in low light (night) conditions showing how different lighting temperatures affect low level noise?

What if 6500K and higher temp bulbs are employed, does this reduce noise more than 5500k, and 3200k?

Or, is 5500k "daylight balanced" the optimal bulb to use with this camera, period?
 
5000K is optimal for the sensor. Daylight is pretty close - tungsten, not so close, hence more noise in the channel getting the least amount of light.

To convert 3200K lighting to 5000K you can use the B-5 gel (may be called something else locally). We've done so with excellent results. Haven't tried Build 16 yet but I suspect that the slightly lower light level with the gel in place will be more than compensated for by the new noise characteristics of the footage. Amazing.... the most film-like digital sensor I've ever seen. Build 16 really is a new camera!

HTH

Paul
 
Why don't you try some extra noise filtering on blue only? Any good color grading software should offer this.

Humans don't see that much detail in blue anyway, so perceived resolution won't suffer very much.

Regards,

Uli
 
5000 k Is also the number I've heard from RED. I use a 80C filter all the time under tungsten to correct for it. Dropping it in is like magic as it corrects the imag.e
 
You must log in or register to reply here.
Back
Top