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 :)

HDRx and twixtor

Y

Yoav Yerushalmi

Well-known member
Joined
Sep 19, 2009
Messages
66
Reaction score
0
Points
0
How 'viable'/crazy would it be to use twixtor to achieve HDRx at 4k/24fps or 5k/24 fps.. ie shoot at a lower framerate and generate the missing frames?
Or even 'slow' the footage in any NLE..

I've seen some impressive slo-mo done, but would normal viewers notice it?
 
Twixtor already takes along time. Throwing a 4K shot in there...man that could take a century haha.

Twixtor can be used to do some impressive stuff. Although unless you really set up your shot correctly and or know how to mask extremely well, and have the time your results can be flawed. Considering it works better with higher frame rates, and the Scarlets limit of 30fps at 4K and 12 at 5K I'd say your best bet is to use it at 3k, 2k and 1K.

Now using it with HDRx would be another animal I'd imagine. From my understanding is that if there is alot of motion in an HDRx shot it can be a little odd looking? Someone would know better than me. But if you blend them together and then do it I think it would be the same as doing it to any other shot.
 
I plan on using twixtor to amplify scarlets 2k overcrank
 
Because it has nothing to do with the resolution or where the material came from. You would use Twixtor in something like FCP, AE, AVID etc. So, by the time the material gets there, it had been debayered and turned into some kind of QT. So, it doesn't matter, if it is material from RED, ALEXA, SONY etc. If your NLE or AE can handle 4k processing, so be it. Just because Red can shoot 4 or 5k, doesn't mean you will be finishing in that resolution. That is why, here we go, 4k vs 2k
 
jake blackstone said:
Because it has nothing to do with the resolution or where the material came from. You would use Twixtor in something like FCP, AE, AVID etc. So, by the time the material gets there, it had been debayered and turned into some kind of QT. So, it doesn't matter, if it is material from RED, ALEXA, SONY etc. If your NLE or AE can handle 4k processing, so be it. But I bet you, most of the time the finish in your NLE or AE will be HD or 2k. That is why, here we go, 4k vs 2k
Click to expand...

Ah, okay. I didn't know that. I just thought that Twixtor might have issues with 4K being as large as it is. Thanks for the info! =)
 
Last edited:
@ Jake

By the way, are there any articles you'd recommend on bayer and debayering? I know what it is, but I don't completely understand the concept of how it works. For example, when you say, "it had been debayering and turned into some kind of QT," I don't understand what that means... I'm only 17, so I have a lot to learn! Thanks! :)
 
No problem. This is a great place to pick up any kind of knowledge on modern cinema tech.
The best person to ask this question, obviously, is Graemme. He is THE man, behind all of the Red bayer pattern science. Hopefully he can chime in...
 
Brian Iannone said:
@ Jake

By the way, are there any articles you'd recommend on bayer and debayering? I know what it is, but I don't completely understand the concept of how it works. For example, when you say, "it had been debayering and turned into some kind of QT," I don't understand what that means... I'm only 17, so I have a lot to learn! Thanks! :)
Click to expand...

Any search online should yield tons of simple explanations. Basically the sensor on your scarlet is an array of little sensors in a grid. Since CMOS sensors can only measure 'amount of light', they are divided into three colors via tiny filters placed on top of them. Some are red, some are blue, and some green. Because the eye is more sensitive to light (luma) instead of color (chroma), and since the eye is best at discerning detail in green, the array is divided into red/green/blue/green arrays. There are double the green sensors.

This is in contrast to other cameras that might split the light in 3, and send it to three different sensors (3CCD -type cameras), or Foveon sensors (sigma makes some) which can actually detect all three colors at the same 'location'.

Ok, so now you've read the sensor, and have all this 'raw' data. But if you were to display it 'as-is', the image would look extremely green. So, math (debayering) is applied. a green sensor is always placed between two red and two blue sensors. So you can guess at what the 'red' and 'blue' values would have been if you could have measured them there. Similarly, you can guess at all the other colors in locations where you couldn't measure them. You could 'average' things, or you could try to be even smarter and detect shapes/patterns. This is known as 'demosaicing' (feel free to look that up on google to learn more).

Alright, so now you have this 'demosaiced' information. It's still in theory a 5k image, but really doesn't have the full 5k resolution, as you have been guessing at a lot of the intermediate pixels. The new canon C300 will instead spit out a 1080p scan from its 4k sensor -- thereby really knowing the full color at each 'pixel'. Red gives you more flexibility, by allowing you to decide what to do with the data.

The last part 'turned into some kind of QT' I assume the poster is just referring to most workflows people employ. You shoot raw, then use something like redcine to 'color grade', and output ProRes (in a QT container) which you can then edit in final cut or avid or premiere or whatever. Prores is your demosaiced/color graded/processed/downrezzed data, and is presumably what you would be applying your effects to.

I should point out that in Premiere 5.5, and FCPX, you "may not need" to use redcine or move out of your 'raw' data at all. I'm guessing as I have not gotten to play around with them, but from what I've read it seems to indicate that it will accept the raw red footage, and (silently in the background) render it to some intermediate (prores) format for you to view as you edit (using LUTs or some other color correcting methods). However, all the real work will still be done on your original footage until you do a final render to your output.

Anyways.. short version. I think you'll learn a lot more by just doing some searches on google...
 
Yoav Yerushalmi said:
Any search online should yield tons of simple explanations. Basically the sensor on your scarlet is an array of little sensors in a grid. Since CMOS sensors can only measure 'amount of light', they are divided into three colors via tiny filters placed on top of them. Some are red, some are blue, and some green. Because the eye is more sensitive to light (luma) instead of color (chroma), and since the eye is best at discerning detail in green, the array is divided into red/green/blue/green arrays. There are double the green sensors.

This is in contrast to other cameras that might split the light in 3, and send it to three different sensors (3CCD -type cameras), or Foveon sensors (sigma makes some) which can actually detect all three colors at the same 'location'.

Ok, so now you've read the sensor, and have all this 'raw' data. But if you were to display it 'as-is', the image would look extremely green. So, math (debayering) is applied. a green sensor is always placed between two red and two blue sensors. So you can guess at what the 'red' and 'blue' values would have been if you could have measured them there. Similarly, you can guess at all the other colors in locations where you couldn't measure them. You could 'average' things, or you could try to be even smarter and detect shapes/patterns. This is known as 'demosaicing' (feel free to look that up on google to learn more).

Alright, so now you have this 'demosaiced' information. It's still in theory a 5k image, but really doesn't have the full 5k resolution, as you have been guessing at a lot of the intermediate pixels. The new canon C300 will instead spit out a 1080p scan from its 4k sensor -- thereby really knowing the full color at each 'pixel'. Red gives you more flexibility, by allowing you to decide what to do with the data.

The last part 'turned into some kind of QT' I assume the poster is just referring to most workflows people employ. You shoot raw, then use something like redcine to 'color grade', and output ProRes (in a QT container) which you can then edit in final cut or avid or premiere or whatever. Prores is your demosaiced/color graded/processed/downrezzed data, and is presumably what you would be applying your effects to.

I should point out that in Premiere 5.5, and FCPX, you "may not need" to use redcine or move out of your 'raw' data at all. I'm guessing as I have not gotten to play around with them, but from what I've read it seems to indicate that it will accept the raw red footage, and (silently in the background) render it to some intermediate (prores) format for you to view as you edit (using LUTs or some other color correcting methods). However, all the real work will still be done on your original footage until you do a final render to your output.

Anyways.. short version. I think you'll learn a lot more by just doing some searches on google...
Click to expand...

Appreciate that post. Cheers Yoav.

Best
Lliam
 
You must log in or register to reply here.
Back
Top