Welcome to our community

Be a part of something great, join today!

  • 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 :)

Run’n Gun S16 zooms, 2K, 3K and why 2.5K is vital

Hans von Sonntag

Well-known member
Joined
Jan 15, 2007
Messages
1,968
Reaction score
0
Points
0
Website
www.hubbertvonsonntag.com
Hi all,

some thoughts on a subject I've posted about several times here and there and that now deserves an own thread: S16mm doc shooting with RED.

S16mm zooms are the best option for documentarian work: They are light weight and don’t deliver the shallow DOF of 35mm lenses, good for single operated handheld work. The DOF shot wide open is still nice but not as challenging. The possible B4 route is not a good option.


No B4 route.

If you evaluate a B4 HD-ENG lens setup on one RED side by side with a fairly modern S16mm zoom on a second RED you instantly see what I am talking about:

1. RED turnes virtually into a dog slow 160 ASA camera because the necessary B4 adaptor adds 1 stop light-loss. For comparison, a F900R has approx. 500 ASA. Pushing 2K footage in REDCine creates much more visible noise than with 4K footage and must be avoided for satisfactory results.

2. Additionally the FVO is cropped due to the nature of the adaptor, meaning a lens that has a wide FOV on a F900 for instance has a considerably more narrow FOV on the RED.

3. The adaptor contains glas and therefore reduces the performance of the B4 lens to a certain extend which has been developed for an entirely different optical system (3CCD prism 2/3”). Also HD ENG lenses tend to have more CA than S16mm zooms.



2K vs 3K concerning S16 lenses.

S16mm zooms cover 2K windowed, but vignette considerably in 3K windowed. The FOV in 2K mode is reduced about 25% because 2K windowed has a smaller projection area than S16mm. To make a S16mm lens fit with the RED and to get the most possible FOV one needs a windowed format between 2K and 3K, lets say 2.5K windowed.

Now, S16mm lenses vignette differently. A Zeiss 9.5 SuperSpeed MKII vignettes more than a Canon 8-64. Thus, a customisable windowing would be great; making sure that a given lens gets the optimal window of the sensor.


Workflow.

Lets assume that we want a 1080p master. Today there are two ways to achieve this with a S16mm lens:

1. Either one shoots 2K, does a center crop for 1080p and loses FOV,

2. or one shoots 3k, gaines FOV, scales the picture until the lens does not vignettes (approx. 2.5k) and down converts to 1080p.

- Shootig 2K windowed has the advantage that no scaling to the debayered pixel is added.

- Shooting 3K windowed has the advantage that the maximum FOV of the lens is used and considerable “flesh” for oversampling is generated for true 1080p resolution which is with a 2K resolution bayer pattern sensor generated picture not possible since 20-30% of the pre-debayer resolution will be lost after debayering.

- Shooting 3K windowed has the downside of displaying a heavily vignetting picture in the EVF, which is fairly distracting for the operator.



Pictures.

To illustrate how much FOV is lost with 2K windowed I have two pictures attached. One shows 2K windowed, and one shows 3K windowed. Both were shot with a Canon 8-64, the classic wide angle zoom in S16mm photography. Please note that the 3K example shows how 3K windowed mode looks in the EVF, distracting I find. Customised windowing would be a huge step towards versitality.


Summary.

2.5K windowed, or even better customisable windowed modes are necessary for convenient, true 1080p, documentarian filmmaking utilising S16mm zooms.

Hans
 
Hans,

Completely agree about getting as much "meat" as possible to oversample a 1080p image from, but bear in mind that the frames you show are 2:1 (by my eye, at least).

Assuming that I haven't made a total cockup of that, you will probably find (again, simply by eye) that the corners of a 16:9 frame at 2K are getting damn close to the edge of that image circle....

EDIT: Well, maybe not that close! Still probably worthwhile... But... If you assume 75% effective RGB resolution after debayer (which seems more than reasonable) then for 1080p output you would ideally want an acquisition resolution of about 2.5K precisely, and I doubt you could hit that target, looking at those images, at least in 16:9 (I make it 2,560 horizontal pixels required for 1920 output @ 75% effective resolution).
/EDIT

Nice idea though, just wanted to check if you'd overlooked that? Also, of course you may want to shoot 2:1, in which case it's a different story!

If you want oversample room - just thinking off the top of my head here - there's an argument for shooting 2.35:1 utilising as much width (now you'd be well over 2.5K, judging by your images above) as possible within the image circle - of course that's only really useful if you don't mind the 'scope ratio for your project, but that's never an issue for me as I love widescreen! Just a thought...
 
Dominic,

chopping off the left and right sides in 3K for 16:9 does not give you a vignette free 3K 16:9 crop. Unfortunaltey. And even if it was you would still have that annoying vignetting in the viewfinder/client monitor.

We need a clean windowed mode for S16mm lenses which does not steal 25% of the possible FOV and oversampling meat.

Hans
 
Call me crazy, but if you are going to lose the 35mm DOF, and inherit the Red's comparatively cumbersome (as of now) work-flow, why not just stick with a Varicam (or others in its class) for this kind of shooting? If you are shooting for television, the only reason I can think of to go with the Red camera is the 35mm DOF -- although that is reason enough for me to have put my order in.
Rob Gardner
 
Call me crazy, but if you are going to lose the 35mm DOF, and inherit the Red's comparatively cumbersome (as of now) work-flow, why not just stick with a Varicam (or others in its class) for this kind of shooting? If you are shooting for television, the only reason I can think of to go with the Red camera is the 35mm DOF -- although that is reason enough for me to have put my order in.
Rob Gardner

good point - sometimes it's easy to overlook the forest when focusing on the tree.
 
Call me crazy, but if you are going to lose the 35mm DOF, and inherit the Red's comparatively cumbersome (as of now) work-flow, why not just stick with a Varicam (or others in its class) for this kind of shooting? If you are shooting for television, the only reason I can think of to go with the Red camera is the 35mm DOF -- although that is reason enough for me to have put my order in.
Rob Gardner

Yes I do agree. There are lot of applications where a 2/3 video camera is better suited, ENG for instance.

But nevertheless there are many like me who like the idea of shooting one part with 35mm primes and 4K and another part with a S16 zoom because it's great for run an gun. Also the more you window the sensor the more fps are possible.

Versatality is key and part of the REDs philosophy.

Hans
 
I don't agree, Red has numerous advantages besides 35mm DOF. Latitude is one. Raw workflow although cumbersome as of yet has it's advantages too, so does an all digital workflow, 6 hours of 2K on one hard drive ! Also if one has decided to invest in a Red camera, that's what he'll be shooting with, adapting it to his different shooting situations or projects !

Cheers,
Damien
 
I think you'd drive yourself a little crazy with the customizable resolutions. Perhaps better would be to simply ask for a Super-16 - equivalent resolution, one that yielded a sensor area the same as that film format. Then you would know that the lenses would all cover properly. I haven't done the math, but I suspect it is around 2.4K or so.
 
onother fininancial reason to consider that might trump varicam is if you already own all s16 stuff but don't have aks invested in B4 system and you also (of course) plan to use red in full res. I't like getting a free s16 cam with every 35mm cam. who would say no to that?
 
I think you'd drive yourself a little crazy with the customizable resolutions. Perhaps better would be to simply ask for a Super-16 - equivalent resolution, one that yielded a sensor area the same as that film format. Then you would know that the lenses would all cover properly. I haven't done the math, but I suspect it is around 2.4K or so.

I'm perfectly happy with a S16mm equivalent windowed sensor. But since RED seems to be basically a computer with an attached sensor I think customisable resolustions are easy to implement via software. Please correct me If I am wrong,

Hans
 
It may be easy, but what a pain to deal with both on set and in post. Consistency breeds efficiency.
 
I'm perfectly happy with a S16mm equivalent windowed sensor. But since RED seems to be basically a computer with an attached sensor I think customisable resolustions are easy to implement via software. Please correct me If I am wrong,

Hans

As I mentionned in another thread, it's probably easy enough on the camera end, put probably not that simple on the debayering end (Redcine, Redalert, Scratch and all the other solutions we all hope will soon become available). So yes, maybe a true S16 resolution would be nice, how much would we really gain ?

Cheers,
Damien
 
It may be easy, but what a pain to deal with both on set and in post. Consistency breeds efficiency.

I would say leave it to the customer to mess things up :)

For my part I would like to see the possibility to ajust a certain lens - say Canon 8-64 - to a windowed mode that would fit to this paticular lens.

Hans
 
As I mentionned in another thread, it's probably easy enough on the camera end, put probably not that simple on the debayering end (Redcine, Redalert, Scratch and all the other solutions we all hope will soon become available). So yes, maybe a true S16 resolution would be nice, how much would we really gain ?

Cheers,
Damien


Damien

I guess approx. 25% more usable FOV. I find ist a pitty to have a nice wide angle 8-64 that becomes in 2K mode a boring 10-80 equivalent (FOV-wise, not in focal length, we don't want to open this can).

Hans
 
I asked for the same thing a while back and Deanan answered this:

"Down sizing to 2k with an odd scale softens things a bit (to varying degrees depending on the algorithm) which can easily negate the benefits of the extra 130 pixels. It also makes things more complicated for the QT proxy resolutions with very strange sizes.

What you may want to do is shoot 3k and crop/scale depending on the set of lenses you will use. Longer lenses will cover more than wides."
 
What about other focal lengths?

What about other focal lengths?

Hans,

About s16 lenses for 3K:

I'm assuming you've got the Canon @8mm... what happens at 12, 16, 25... less vignetting? I would assume so...??? also, what stop are you at... I'm assuming this would also have an effect...???

curious since we have plenty of s16 lenses sitting around.

Jack

Red twothousandthreehundredsomething
 
Hans,

About s16 lenses for 3K:

I'm assuming you've got the Canon @8mm... what happens at 12, 16, 25... less vignetting? I would assume so...??? also, what stop are you at... I'm assuming this would also have an effect...???

curious since we have plenty of s16 lenses sitting around.

Jack

Red twothousandthreehundredsomething

Nothing, it vignettes in all focal lengths almost the same. T 2.4, open wide. Didn't stop down, since I don't expect anything particular changing and the lens must work wide open anyway.

Hans
 
Back
Top