If you put 4K footage into a 1080 HD comp, it will come in at 100% scale. The layer will be a LOT bigger than than the comp itself. You'll have a precise 1-to-1 pixel correspondence with the comp itself. Thus, you can simulate pans & tilts easily. You can recompose shots. You can set a keyframe at 100% scale and simulate a zoom in with no worries about image degredation.
And if you choose to shoot green screen of a person standing & talking to the 4K camera, you're in real luck: you can turn the camera on its side, and just use it in AE by rotating it 90 degrees in the proper direction. You can take advantage of more pixels: the subject is vertical, so why waste all those extra pixels by keeping the camera in its normal horizontal orientation?
I'd imagine you could fake a zoom in by scaling the clip from full-figure to a head & shoulders shot without going beyond 100% scale in the footage. I used this particular trick more than once when I produced spots in SD, but shot in HD. It works like a charm.
This is exactly the info I was looking for. Thank you. There seems to be more "Pro's" than "Con's" around the 4K camera. For us, the biggest bonus, will be the panning, tilting and zooming choices it provides.
Thanks again, and keep an eye out - we are starting production proper in a month or two! Hold thumbs!!
This might be an "unfair" question to post here in the forum, (but I am going to do it anyway...)
Knowing nothing about 4K cameras, but finding myself highly interrested in them, in the experience of the forum members, what models am I looking at?
I dont even know where to start looking. The camera we need will have to fall in the BBC approved camera list, I have the list, but the names all look the same to me!
I have been tasked with finding a suitable camera and a suitable cost to see if it is going to be possible to pick one up before the sale of Season one (and the ultimate world domination planned after that! ) - of course, once season one has been sold, we might just buy the camera manufacturers and have them make a new one to our specifications! sigh, sorry, I blanked out there for a sec..... where was I...
Oh yes, I am going to go through the list of approved cameras and look up their specs to find the 4K ones, but any advice would, as usual, be highly valued and appreciated.
after a day of reading, I am also really interrested in getting know more about "Snorkel" and Periscope" lenses for really getting into those miniature scale sets....
I don't know much about 4K cameras but I know the Red Scarlet seems to be a popular choice. You can get a model that works with Canon lenses. It's also fairly compact so it's easy to set up and break down. I think Sony has a nice one coming out soon as well.
It's only 2K, but I saw a demo of the ARRI Alexa when it came out and I definitely think it's worth looking at if you have the money. Of course it all depends on your budget. A more affordable 2K option is the Blackmagic Design Cinema Camera. Otherwise the Red Scarlet or the Red Epic are both good 4K choices over the Red One especially if you need to shoot low light.
If we do get one of the RED cameras, is it still worth porting the RAW data out to a recording device, like the Samurai, or would that actually be "dumming" the footage down?
It depends on what you intend to do with the footage. If you don't intend to do much with color, then going straight to ProRes or DNxHD and baking in the color settings is no big loss. Same with transformations and going straight to an HD capture from a camera that can capture a larger frame. It's all a matter of tradeoffs, as usual.
As Todd says, it's always about tradeoffs. Christopher Nolan shoot as much of "Dark Knight Rises" on IMAX cameras using real sets and real actors. Much of the footage of the opening airplane capture was shot for real and only enhanced with visual effects because he had the budget and the crew to make it better. The big Semi Truck flip was a real truck and a real stunt because Nolan minimized the effects and maximized the reality. The difference between the effects in Dark Knight and something like Transformers (a ton of blue screen and CGI) is amazing to my eye but to some it isn't that significant. If you have time take a look at the 'work' that went into the football scene. Way more than was needed but a much better product in the end.
Thanks for the response. I was really interrested to see if there are any specific reasons we owuld use the Samurai with a RED camera. We use the Samurai with our current AF101, because the native footage from the camera do not meet the BBC standards. I assume that a camera as powerful as the RED would have no hassles with this though, so we would probably just use the footage directly from the camera.
I am now sitting with the question of Red One VS Red Scarlet for shooting miniatures. We have a DIT coming in hopefully in a few days to give us the run down in the differences, anddo some test shots to see what would suite us better.
just an update (kind of). I have joined the Red forum, and recieved a brief from our producer. Ill add it here, it gives a bit more info on the scales etc we are working with. Once again, not really an AE question, but I hope others will find it interrsting, and Ill add the updates from the Red forum. Hopefully others will gain some advice from it!
2 months and counting till production starts....... (hold thumbs guys)
" We are in pre-production for a children's show for HD output, 1080P. We currently have a Panasonic AF101 shooting 1080, with external recorder, which can achieve approximately 75% of the shots we would like. However we think a RED can facilitate some of the moving shots we would otherwise find very difficult, in the context of this project.The show involves small aliens, approximately 20cm tall (approx. 1:6 scale compared to humans), with action taking place on spaceships and on earth. We are using real puppets for the characters - actual size, i.e. 20cm tall, using chroma keying rods for their motor movements, e.g. walking, arm moves, and then adding animation details, e.g. speaking mouths, facial expressions, finger movements, in post.
We then also have a number of miniature sets and miniature ships, of varying scales. Depending on the shot, we will either shoot the characters against blue screen and composite in, or actually puppeteer them in the set.
Specifically then, the reasons we are considering a RED:
1. Simulate camera motion in post -A large proportion of shots involves several of the characters being composited into a set / scene. Because we can only puppeteer / make one character move at a time, primarily using chroma keying rods attached to the characters, we’d have to make multiple passes, one for each character to make the final composition. If we don’t want any camera motion in the shot, this is no problem, we just use the standard 1080P camera in a fixed position and capture the several takes. However if we want any pans or zooms, because the camera isn’t motorized, we can’t replicate the movement exactly, and hence we can’t make the composition. So the 4k would allow us to generate the moving 1080 image from within the stationery 4k space. This movement is important to us simply because we want to be able to maximize the cinematic value of the show.
2. Easier to handle motion in post than camera side -
Even though we can execute some of the shots with the AF101, because the characters are approximately 1:6 scale any unevenness in camera motion gets exaggerated. So if we wanted to zoom, pan or rack between two characters it would seem, we’re thinking, easier to add that motion in post (maybe racking camera side). One drawback is perhaps the loss of parallax, but we’d mix it up where we can with actual dolly moves / pans.
3. Framing / Macro shots -
4k images will give us framing options in the aesthetic of shots, hopefully reducing re-shoot time. We also have a number of macro shots planned. We could get very good images with the right lens on the AF101, but I’m thinking the 4k images gives us greater control with the DOF, in that we have a margin to move physically away from the talent, especially as a lot of shots are relatively low light, and we’re struggling with too shallow a DOF. Is the Red Scarlet’s shallow DOF a problem here,
4. Time consideration -
This is initially being pitched as a 13 episode project, so we’re looking for longer term time efficiency to get the shots we’re after. Simply put, we’re thinking the degree of options / control of framing the shots in post is a quicker method that capturing the exact shots initially in 1080P, and then having no room to maneuver.
1. Shallow Depth of Field - although there are several shots where we want a shallow DOF, there are many more where we need it deeper, but is made more complicated but deliberate low light for mood / contrast.
Below is a basic plan layout of the principal set, the bridge. The set is approx. 80 x 100cm. With lighting, it's relatively low IRE overall, with High lumen LED lighting on the characters, low lighting on side walls, and deliberate shadows in the corners / back wall.
1.1 Can we achieve a 60cm or greater DOF in this low light miniature set, to get all the characters in focus?1.2 Can we achieve a 20cm or less DOF in this set? So as to get singular characters in focus and rack between them?
1.3 Which len(s)?2. No high fps - We usually have the characters' final running speed at 25% recorded - we shoot at 50 fps (UK broadcast), so we achieve 50% speed no problem, and then have to add artificial frames again in After Effects to get another 50%. This is also the case with the model spaceships. Would have been nice to have it higher, as the added frames aren't always perfect.3. Learning time
4. Delay in shooting (a few weeks)
5. Storage space
6. Cost - we can employ someone for a year instead!
So, the key questions are:
1. Is a RED suitable for what we're asking for here?
2. Is there a better solution?
3. If we did opt for the RED, what are the best lens for the shots we require – the macro shots, DOF needs, general purpose shooting lens? We would clearly prefer stills lens from a costings perspective. Other considerations is that it's for HD TV output (not cinema but neither SD); that we're shooting a lot of miniatures; that we would like to focus rack between miniatures.
4. What other accessories would help us to maximise the RED's use in this project "
Here are some lens and format considerations you might not have thought of. These are basic principals that apply to all photographic systems.
- Perspective is controlled by camera position. The common misconception is that a wide angle lens gives you different perspective than a telephoto. This is not the case. A wide angle lens gives you a wider angle of view, but the perspective, the relationship between foreground and background, and the apparent movement between foreground and background objects is exactly the same with any lens if the camera position is the same. Think of focal length as a cropping tool.
- The larger the image sensor the more shallow the depth of field. An IMAX camera has a very large image area. A 35mm film camera has an image are about 1/8 the size. A 16mm film camera has an image area about 1/4 the size of a 35mm film camera. Put a lens on each of these cameras that gives you say a 40º angle of view, set all of the apertures to f5.6 and the depth of field with the IMAX camera at a focus distance of 1 meter will be a few millimeters, the depth of field with the 35mm camera will be a few centimeters, and the depth of field with a 16mm camera will be pretty close to a meter. Shoot the same scene with a typical smart phone camera and the depth of field will be several meters. It makes no difference what kind of sensor you have, the larger the sensor, the shorter the depth of field for the same angle of view and f-stop.
- Sharpness or clarity is not 100% dependant on sensor size. The amount of resolution you can get from an image sensor or a sheet if film comes from the physical structure. Extremely small image sensors can be manufactured with amazing resolving power. The better (means more expensive most of the time) the quality of manufacture the better the image.
Why am I giving you this information. When shooting miniatures depth of field is always a problem. In many cases you are trying to simulate full size sets. This means more light for a smaller f-stop or smaller image sensors. When shooting film there is a limit of how much you can do with a 16mm negative so most minature work in film has been done with 35mm or even 65mm cameras. With electronic cameras, especially good ones, you can get amazing resolution from smaller image sensors. If it were my project I would stay away from large sensor cameras. A good 2/3 inch camera will give you the same approximate sensor size as a 16mm camera while a Red or an Epic will have a sensor that is close to 24mm so it will give you about the same depth of field as a 35mm film camera. For the same depth of field the Red is going to require twice as much light as a 2/3 inch camera.
Another problem you face shooting miniatures is that you have to scale the camera position like you scale the miniatures. What do I mean by that? Typically, when I'm shooting a dialogue scene between two actors I'll have the camera about 3 feet or 1 meter or less from the actors. This perspective is similar to what you get when you are talking with someone face to face. The perspective of the close camera invites us into the actors world and makes us more a part of the conversation. If you want the scene a little more detached you move back and shoot with a longer lens. If you want the scene more personal, you move in.
Now consider the same problem when your minature set is 1/4 scale. If you want the same perspective, the same feeling of intimacy, then your camera position isn't going to be a meter away from the puppet, it's going to be 25cm. This means it's going to be very difficult to get a big camera in the scene to achieve the same perspective. This also throws a whole new set of problems with depth of field. The closer the focus point the shallower the depth of field. If f 4 gives you what you want for a look at 1 meter, then you're going to need something like f 11 to get that same depth of field at 25cm. Change the image sensor size from 24mm (typical large or full sensor size) to 2/3" (typical professional HD video camera) and then a shot from 1 meter with a full sensor camera at f4 will have about the same depth of field as a shot with a 2/3 or 1/2" sensor from 25cm.
I once produced a training film for Weyerhaeuser Paper company where we used a big minature railroad set. The only way we could get a camera in and around that minature set and achieve the perspective that I wanted was to use a Kenworthy Snorkel. We also had to use 2K and 4K HMI's to light the set because there was so much light loss through the snorkel and I wanted as much depth of field as I could get. Here's a shot of a snorkel system on a mineature set and this is how they work.
I hope this gives you something to think about.
The common misconception is that a wide angle lens gives you different perspective than a telephoto. This is not the case... Think of focal length as a cropping tool.
I tend to agree with people who share that misconception, and here is a brief example why.
Shot taken with 50 mm lens:
Shot taken with 35 mm lens from the same distance:
Here is the overlay (never succeeded in matching):
What am I missing?
If you have the exact same camera position with a perfect lens = exact same perspective. Light not passing through something or near something that bends it travels in a straight line. The miss-alignment you're seeing is lens distortion. Shoot the same two shots with a set of Zeiss Primes and the match between a 35 and a 50 will be much much closer. Few pro's will cough up the nearly $4K per lens for these amazing lenses but many, included yours truly, will rent them when there's enough budget in the production. Even the difference between a consumer and a professional Canon or Nikon lens is significant. If you use the Optics Compensation to correct the barrel distortion in your 35mm lens you should get a nearly perfect overlay in the 5 minutes it took me to do this with your shots.
By the way, the first clue that there was lens distortion is that the line created by the edge of the refrigerator door is not straight in either shot. Sorry your lenses are not perfect.
Pretty straight... at 800% I see about 2 pixels of distortion on that edge. Not bad for a consumer lens.
Hi Rick, Fuzzy...
The strangest thing happened to me yesterday ( I am tempted to start with "A crazy thing happened on the way to the library....) <--But I wont!
Anyway, We are doing all the tests now around 4K images and the options it gives us in post etc. Since we do not have a 4K capable camera yet, we decided to take an existing piece of footage, shot on the AF101, import into AE, create a new 4096X2304 Comp, import the 1080 footage and "transform", "fit to comp".
The 1080 footage, scaled up to the larger comp, has extremely little difference. Now I am not sure if i was drinking something other than water yesterday.... but I thought it was pretty amazing.
Here is a screenshot of one of the characters eyes, first in the native 1080comp, then transformed to fit in the 4096X2304 comp
There is a definite softening in the image, but it really seems to me, to be a small difference?
If this is the case, what stops us from using our existing equipment to shoot some "larger" shots to allow for extra post options?
(I have a feeling I am making a terrible fool of myself with this question, and I am missing something very obvious! If thats the case - I apologize upfront! )
I have a feeling I am making a terrible fool of myself with this question, and I am missing something very obvious!
What you're probably missing is the following: 4096 x 2304 is 2.133 times larger than 1920 x 1080. So as to get the same object size inside 1920 x 1080 frame (composition) while shooting 4k you should move your camera away accordingly. Try shooting your puppet from both original distance and this newly calculated one with Full HD frame, import footages into AE, drop the latter into 4k comp, scale to fit, drop 4k comp into 1080 comp and compare the result side by side at 100% view. Here is what I've got (guess which one is which):
The shot on the right was taken from farther away. Even if the resolution was as good as he one on the left the perspective (notice the width of the shadow and the curve on the top of the battery) is different. See my comments on camera position and perspective.
With the kind of detail that is in Pierre's sample images you can probably get away with some scaling up. You'll get even better results with 3rd party software like Magic Bullet's Instant HD which now includes Digital Anarchy's Resizer.
Pierre, you said you "fit" the footage to the 4K comp. Using AE's "Fit to Comp" is a BAD IDEA... Fit horizontally or fit vertically. When placing scaling footage it is imperative that you have the X and Y scale values exactly identical. Even if one shot has rectangular pixels and the comp is square, never FIT video to a comp frame. Always Fit Horizontally (Shift + Ctrl/Cmnd + Alt/Option + h) or fit vertically to a comp, or just scale the footage. I didn't see much if any distortion in your samples so I'm assuming you just scaled up your footage.
Having a little extra room when shooting miniatures is a good thing. In all of the work that I have done I have never had the need to have 4X the room in a shot that you would get with a 4K camera. The big advantage of a 4K camera is that you have access to more data for color correction and other effects like keying. You will get much better edges.
Speaking of edges, keyed edges go bad really quickly when you scale up footage. If you end up doing that then it's imperative that you scale up and pre-comp or render before you key.
I know you're pretty set on using green screen and I wholeheartedly understand why so many people prefer it over other background replacement methods, but I still think there's something to be said for using a rear projection screen with a good short-throw projector (or HD projector if you have the budget and room) and doing your background replacement in camera. Of course I've always preferred in-camera techniques to hours of post-production work and part of me likes rearscreen projection simply because it's a dying art, but I think you might be surprised at how good it can look. In the industry they call it the poorman's process, but I think if it's done right it can look great and it might actaully end up saving you a lot of time and money on the post production side of things.
As of late, I've been using rear projection for a bunch of animated and live action projects, and although it has a very unique look, I think it can work wonders for many scenarios. Here's an example of an experimental animated/live action film in which we used a rear screen projection for all the stop-motion scenes of the rabbits. And, because we weren't keying we were able to get perfect definition on the edges of the rabbit fur, something that would be a lot harder if we were using green screen.
Since there was no depth in the background projection, which was just a digital photograph of trees, we added some branches in the foreground and shot through a piece of glass that had some hand-painted grass on it. We did both of the dolly shots using a huge motion-control rig, which you probably won't have access to, but if you decide to shoot any of your miniatures in stop-motion there are alternatives to traditional motion control like moco camera sliders. I say this because if you have the time, you might consider shooting some shots in stop-motion using moco as an alternative to live action, because you'll be able to shoot at whatever f-stop you like simply by compensating with a long exposure. You can also repeat camera moves, doing different passes for all the lights to composite later and you can do all this with a simple DSLR (which will shoot 4K or higher in stop-motion) and some frame grabbing software like Dragon Frame. Another way of getting perfect mattes in stop-motion is to use a front light / back light technique. I realize this could potentially change your entire budget and approach, but I think it could definitely be useful for a few key shots where you really need the the depth of field. Of course, everything depends on your aesthetic. Let me know if you're interested in any of these suggestions and I can give you more info on how to do this stuff.
Thanks for the reply. We actually want the 4K image to be used in the smaller 1080p viewing space. Specifcally so we can add post pans and zooms. It just seems to add a little extra otions for us. I see the image distortion when scaling though, i just thought that it would be a LOT worse than it seems to be.
Thanks for the advice. I used the 'Fit to Comp" in this instance only to provide a (lets call it a proxy) proxy piece of footage at 4K resolution to test the post panniong and zooming. I was not aware of it being a bad idea - thanks for the heads-up. The whole thing was just surprising to me, as I assumed that the quality would be almost unusable after the scaling, but even with the distortion and slight blurry edges, it really is not THAT bad?
Thanks for all the info. Ill run it by the team, and give feedback.