High ISO - image looks clean, playback looks noisy

[David Battistella]

Anton, I don't mean to answer for graeme, but I have found these modes most useful for stills capture. What the averaging does is expose the number of frames, it then rapidly processes those frames into one frame and writes that to the r3d. It takes the best of the exposures to combine into one exposure. For stills if you average 16 frames its making 16 exposures and combining it into one exposure. it takes about 32 seconds to average a frame this way but the results are very stunning.

If you are making table top images for example it's sampling the frames and drawing one perfect exposure/noise profile frame out of that. In motion (on my EPIC dragon with the vintage REDMAG) I can grab 24P material by setting the camera to 24fps and recording with a two frame average. You might be able to do burst recordings with 4 frame or 8 frame averaging but your ability to record more than a few seconds would be limited.

The advantage to these modes is that it is processing two exposures and averaging (and presumably weighing) the best parts of the exposure to the frame. If you get into a low light situation then what this means is that it will average to and eliminate noise (let's say) because it can take the best of both exposure to combine into one frame. Sorry if it is not more detailed but it has been about 8 months since I did my initial tests.

As to the dropping the ISO, you will see a drop in exposure when you go into averaging mode, like if you go from 24 to 48 fps and spending on whether you are set to absolute or relative shutter modes. When you 2 frame average in motion you set the camera frame rate to what you want the recorded frame rate to be. IE 24FPS and then the camera will record at 48 and do the averaging for you. It's a cool concept because you are using 2 exposures (2 frames) to create the final exposure. In the case of stills you can use up to 16.

David

 

[Anton Shavlik]

Thanks for the thorough response, David. I understand it will mostly be used for stills and timelapse in practice, but I imagine if it has advantages for stills it will have advantages for motion as well, as some others have said in this thread.

It's also been a while since I played with the feature so I just ran some tests myself. I am really curious what the strengths of this technique are, since averaging can rid some of the noise but requires a higher ISO, and how it may affect your motion blur.

No changes made to the lens between the shots,
Top: 5k 96fps 4:1, 4 frame average, 3200 ISO, 180 degree shutter
Bottom: 5k 24fps 4:1, 800 ISO, 180 degree shutter

Motion blur:

The key difference for noise is that I blackshaded for 1/200 when shooting 96fps and 1/48 when shooting 24fps. Both my T and E were green and the two shots were taken about a minute apart, but it's possible one blackshade is better at crushing the noise than the other. Still, visually I don't see much advantage. Obviously averaging 4 frames does give a 'strobbing' effect in your motion blur.

 

[Anton Shavlik]

Another thing to note is that while averaging, my Red 7" monitor was running at about 12fps but the resulting footage was indeed 24fps. Clearly, it takes a lot of processing power to average 4 frames before compression.

 

[David Battistella]

Thanks for sharing these frames. The big advantage comes with stills of TL work for sure. I don't like to mess too much with motion blur as its such an important component of visual literacy.

David