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 :)

Added Cost In Post of Shooting Anamorphic

M

Matt W.

Well-known member
Joined
Oct 3, 2009
Messages
1,350
Reaction score
1
Points
38
Just curious what the added cost in post of shooting anamorphic is. I assume most of the cost is on set (more expensive camera to get a larger sensor–or more film consumed, more light to get stop, more tricky work with diopters, etc.) but working with anamorphic footage in post has been a headache for me. Lots of desqueezing, some tricky color correction, more difficult to track and motion match.

Am I just inexperienced with this or is anamorphic a lot harder (and therefore more expensive) to work with in post, too?

I love the look but maybe not enough to justify it.
 
I don't find anamorphic any more expensive unless you need to 3D camera track or comp out lens flares. All of the headaches are on the 3D side with matching elements not simple post-production like color and editing.
 
Yeah, it's probably my inexperience. But the lens flares are a huge pain because I have to paint them out, stabilize, then paint them back in, and Boujou doesn't love all the distortion. Hard to patch elements in in mocha because After Effects crops the edges when you do lens distortion–>track–>undistort and so you have to make oversized comps (at least that's my approach) then mask out the edges in mocha so it doesn't try to track them.

Do you have any recommendation for workflows?

To be fair, this is my second anamorphic project ever. And the first one had to be outsourced because I couldn't get the tracks right (then again, neither could the very high end post house that did most of the effects on their first few tries).

In 3D it sounds like a nightmare.
 
Roughly speaking and not scientific - too tired today :cheers2:


Anamorphic doesn't require expensive cameras/large sensors.
Sure, sensor size plays a factor in what anamorphic lens you use

Take an alexa, a crop dslr, or even an iPhone,
use an anamorphic lens (or clip/clamp-on lens ascessory)
shoot in 4:3 mode and or crop, then desqueeze(scale) to a 2.4:1 presentation

Desqueeze is just a horizontal or vertical scale - thats it
(although some prefer to change the pixel shape)



For post costs, it depends on how the de-squeeze is handled.
And at what point in the process it is.


Side windowed acquisition squeezed down for 6K Post files, maybe cropped to 5K

OR

6K actual stretched horizontally for 8 to 12k Post files (depending on anamorphic lens factor)


The latter is more processor/storage heavy and should cost more
but man, 6K streched to 8 to 12K, results in super sweet 4K


The main point to anamorphic photography is:
Wider FOV wider in relation to shallower DOF
IE, basically you can get the DOF of say a 50 w/ the FOV of a 35

Horizontal streaks and flares are a side effect/benefit/detriment per your perspective


So if you want a similar 74(ish) degree Angle of View of an 18mm normal spherical lens on S35,
then, If you are using a 2X anamorphic lens set, you would select the 35mm lens.
By doing this you get the shallower DOF of the 35, with the wider FOV of the 18


So knowing this,
if you want to use a 20(something) 2X anamorphic lens,
be prepared to light/dress the set for fisheye-like coverage.


Also any post compositing/stabilzation/roto work may be more time consuming and problematic due to flares and image quirks specific to anamorphics, and thus will cost more for more time/work needed
 
Daniel Reed said:
Roughly speaking and not scientific - too tired today :cheers2:


Anamorphic doesn't require expensive cameras/large sensors.
Sure, sensor size plays a factor in what anamorphic lens you use

Take an alexa, a crop dslr, or even an iPhone,
use an anamorphic lens (or clip/clamp-on lens ascessory)
shoot in 4:3 mode and or crop, then desqueeze(scale) to a 2.4:1 presentation

Desqueeze is just a horizontal or vertical scale - thats it
(although some prefer to change the pixel shape)



For post costs, it depends on how the de-squeeze is handled.
And at what point in the process it is.


Side windowed acquisition squeezed down for 6K Post files, maybe cropped to 5K

OR

6K actual stretched horizontally for 8 to 12k Post files (depending on anamorphic lens factor)


The latter is more processor/storage heavy and should cost more
but man, 6K streched to 8 to 12K, results in super sweet 4K


The main point to anamorphic photography is:
Wider FOV wider in relation to shallower DOF
IE, basically you can get the DOF of say a 50 w/ the FOV of a 35

Horizontal streaks and flares are a side effect/benefit/detriment per your perspective


So if you want a similar 74(ish) degree Angle of View of an 18mm normal spherical lens on S35,
then, If you are using a 2X anamorphic lens set, you would select the 35mm lens.
By doing this you get the shallower DOF of the 35, with the wider FOV of the 18


So knowing this,
if you want to use a 20(something) 2X anamorphic lens,
be prepared to light/dress the set for fisheye-like coverage.
Click to expand...

How do you get a true wide angle without a 4:3 super 35mm sensor? I know there are anamorphic shows on 16:9 chips and 2:1 chips, but I've never liked the result once cropped to 2.4:1. Doesn't the smaller chip negate the shallower depth of field characteristics you mentioned (though of course the stretched bokeh is more important to me, but I suppose not to everyone, as you mentioned)? Forgive me if I'm wrong, I'm just confused about the whole thing.

Fair enough, though. This is my first go with the format. I love the look but find post difficult, as I mentioned above.

If you have any advice on working with Boujou and Mocha on heavily distorted footage I'd appreciate it. My hacked together workflow is requiring I render out a lot of weird proxies and it's slowing things down unduly.
 
16:9 chips
crop, then stretch. or shoot in 4:3 or 5:4 modes (crop in camera) then stretch

yes, smaller sensors generally equal less DOF
(depending on the project, this may be more or less desirable)

If you don't want the shallower DOF with a similar Angle of View that anamorphic lenses provide,
Just use a regular 18 to 25 wide spherical - which is plenty wide on a 6K dragon!
an 18 spherical, on dragon 6K, may technically be considered an ultra-wide, nearing 100 degree angle of view.

OR (and also if you don't want the added DOF)

adding a mild horizontal blur in post and gaffing fishing line in front of a spherical lens, can get you similar anamorphic look.
There's also a pre-made filter you can buy for use with sphericals: http://www.vid-atlantic.com/products/cinemorph

Matt W. said:
How do you get a true wide angle without a 4:3 super 35mm sensor? I know there are anamorphic shows on 16:9 chips and 2:1 chips, but I've never liked the result once cropped to 2.4:1. Doesn't the smaller chip negate the depth of field characteristics you mentioned? Forgive me if I'm wrong, I'm just confused about the whole thing.

Fair enough, though. This is my first go with the format. I love the look but find post difficult, as I mentioned above.
Click to expand...
 
Last edited:
We already shot in anamorphic. I guess some of that is over my head, though. What I'm asking is don't you need a 4:3 super35mm chip to get the full field of view of a wide 2x stretch anamorphic lens?

Although right now I'm just looking for tips on Boujou and Mocha workflows for anamorphic footage with heavy distortion. So far the post experience has been a lot more time-intensive than working with flat footage, particularly patching flares, set extensions, and motion matching–but again this is my first anamorphic show (second in post, and the first one there was tricky too, lots of painting out flares, and lots of solve errors on the motion matching).
 
Matt W. said:
What I'm asking is don't you need a 4:3 super35mm chip to get the full field of view of a wide 2x stretch anamorphic lens?
Click to expand...

No.


For matching old-school 4:3 2x anamorphic FOV


use a 4:3 S35 sensor

OR

use a larger sensor and window down to 4:3 S35 in camera

OR

use a larger sensor and Crop to S35 in post

OR

use a focal reducer on a smaller chip to match or be larger then S35 AND window in camera or crop in post as needed


Then

desqueeze/scale in post


2x = 200% post horizontal scale up
1.33x = 133.333~% post horizontal scale up

Some folks prefer to vertically scale down


Using larger then S35 4:3 creates more FOV, which you can cut down to match
 
Last edited:
If talking about matching DOF (how much is in focus), then maybe.

larger sensor, less in focus

small sensor, more in focus

a S35 height matched sensor that windows on the left and right sides to get 4:3 would be a near exact match (aka helium)
 
It sounds like most of your issues are with higher end complex tracking/comp work.

From my understanding,
many use nuke instead of aftereffects; however,
I don't know how the boujou move matching program fits into your workflow (never used it)

On a side note, I think this points to a definate problem with many shooting w/ less planing then expecting post folks to "fix it".
 
Thanks, I might have to look into Nuke's distortion correction for future projects. So far my workflows with After Effects have been convoluted. The lens stuff is over my head to be honest, but it's good to know you don't need a 4:3 sensor to get the most out of 2:1 anamorphic anymore. When I've used them before I had to crop and the resolution loss was unacceptable.

I'm using Boujou to get 3D solves for particle systems within After Effects, pretty basic use. I'm a beginner Boujou user and haven't explored its lens distortion options in depth. I think for my personal projects I might just skip out on anamorphic lenses. Not sure why this is proving so much more challenging.
 
Why not unsqueeze the VFX shots to where you want and then create comps/proxies to apply your effects and such, then edit those back in with whatever high end codec you like? The other thing might be to unsqueeze the footage and apply your effects, nest all the effects and then resqueeze the footage back and squeeze the effects accordingly? For tracking shots, I would assume the best thing to do is the first suggestion as you wouldn't have to worry about squeezing or unsqueezing anything and it might help with image tracking.
 
Zack Birlew said:
Why not unsqueeze the VFX shots to where you want and then create comps/proxies to apply your effects and such, then edit those back in with whatever high end codec you like?
Click to expand...
This is what I have seen done at ILM and Cinesite on anamorphic film jobs. I would recommend to the o.p. that they hire an experienced VFX supervisor to coordinate this workflow according to reasonable industry standards -- it's not that expensive or time-consuming to do. The VES Handbook of Visual Effects: Industry Standard VFX Practices and Procedures has a lot on the subject (900 pages), specifically including file types, aspect ratios, conforming, and other factors.
 
Marc Wielage said:
This is what I have seen done at ILM and Cinesite on anamorphic film jobs. I would recommend to the o.p. that they hire an experienced VFX supervisor to coordinate this workflow according to reasonable industry standards -- it's not that expensive or time-consuming to do. The VES Handbook of Visual Effects: Industry Standard VFX Practices and Procedures has a lot on the subject (900 pages), specifically including file types, aspect ratios, conforming, and other factors.
Click to expand...

That's the workflow I'd been using, but I'm trying to avoid all the extra proxies and stuff. So far I've been able to, for the most part. In the future I'll stick to spherical. I'm trying to do a lot of comp and stabilization work myself without having much experience with anamorphic workflows and everything is just a little slower and less precise. Harder to get a good track, more flares to paint out, etc.

For someone with as little experience as myself, anamorphic definitely does add time to post. For others, I suppose they have the experience to work with it as easily as with spherical. I feel sort of foolish, but I think the take away for me is just realizing my limitations as an artist and the limitations I have in terms of putting up with kludgy workflows.

I'll definitely start training in Nuke if that makes this thing a lot easier. Even if all that does is obviate the need for destretched proxies when going to mocha and boujou it'll save time for the toughest shots and add accuracy to the easy ones where I would normally try to avoid rendering additional proxies for simple paint work, etc.
 
Matt W. said:
That's the workflow I'd been using, but I'm trying to avoid all the extra proxies and stuff. So far I've been able to, for the most part. In the future I'll stick to spherical. I'm trying to do a lot of comp and stabilization work myself without having much experience with anamorphic workflows and everything is just a little slower and less precise. Harder to get a good track, more flares to paint out, etc.
Click to expand...
If you were working with a true squeezed anamorphic image, it'd be a huge chore trying to (say) create digital backgrounds or digital characters and place them in the shot, because now they have a built-in squeeze. If you're looking at a letterboxed image where everything is of normal size, then it's not a problem drawing a circle or a square and have it be accurate to the real shape it is in real life.

There are a lot of other good books on VFX, but the VES Handbook is a good one to start with. They also have some pretty good ones on VFX production and VFX budgeting, which I think are by far the hardest things in the whole business. I have seen shots that started out at $20,000, then had five changes, and by the time the dust settled the company had redone them 100 times, to the point where they lost a lot of money on them. Now, imagine having 100 shots or 200 shots like that in the film. It's not hard to see why so many VFX houses have gone belly-up in the last 10-15 years.
 
For simple stuff, I was just changing the pixel aspect ratio of the footage in After Effects, as one would when working with NTSC or DVCPROHD. For complicated shots, unstretched plates. For more complicated, unstretch, undistort, denoise, flare paint out, etc.

This is just a personal project and I'm doing (so far) all he work myself. It's not that the shots are complicated, most are basic. But even stabilization and paint out is made a lot harder by flares and distortion, for instance. My particle systems don't quite look right because the anamorphic bokeh is so distinctive and After Effects' basic lens distortion correction tools don't work with complex distortion. Mostly it's the flares, though. To stabilize a shot is harder in the first place, but then I need to paint the flares out and paint them back in. The streak flares are hard. The other flares have a distinctive look that's hard to emulate and they bounce if you just stabilize without painting them out and painting them back in. I've been living with that for really long takes and gimbal/steadicam shots, but if I were to do this properly it would be a pretty involved step.

I guess most of the posters here are used to working on much higher budget work where flare paint outs and de-squeezing aren't big deal and are pretty inconsequential in price/man hours compared with ILM style magic. I should have stated that I was just working on a personal project and I usually work on lower end stuff. I make a living as a sup and senior (usually lead) comp artist, but usually on productions with total budgets in the six to seven figure range, and on five-figure contracts for myself, and usually simple stuff shot spherically on an Alexa, so really really easy basic work that anyone could do if they could deal with the tedium.

For inexperienced guys or one man bands, I do think this stuff is trickier with anamorphic. Even with desequeezed plates I found getting a camera solve extremely difficult and know that the high end (very high end) post house we were interfacing with struggled equally. Flares covering faces can be so difficult, because when someone talks there's no way to patch that without a lot of mesh warp. Repainting a face is way outside my skill set. But even stuff like the lens correction tool in After Effects requiring making a bigger comp size because it pushes stuff out into the edges, then masking those areas out on Mocha and Boujou, bringing those elements in, re-distorting them, painting flares back, etc. It's just not as simple or clean anymore. I'm grateful for the recommendation of switching to Nuke to obviate this problems, which I'm going to start training on ASAP.

Budgeting has never been my strength in anything. That's a whole other story. Scope creep is the worst. I know I'm bad at that stuff and pick work carefully. Probably holds me back.
 
Last edited:
Arri Master Prime Anamorphics are a breeze doing vfx, also they only flare when you want them to. There is basically no lens distortion using MPA's, unlike older anamorphics. Another nice thing is all the file sizes are basically in half. If your ok in nuke/maya, the vfx render cost is about half since your only doing half the pixels (i.e. it is fairly cost effective to do 2048x1716 - 4k widescreen height for vfx, while normal 4k vfx gets too expensive for most productions).
 
Well, that settles it... going to learn Nuke. Seems like it's the way to go.

And agreed on 4k vfx. I'm glad I'm not the only who feels that way. The render times and ram requirements add up. That said, those lenses aren't ever in my budget and the distorted look of anamorphics is what appeals to me visually if not technically.
 
Matt W. said:
Well, that settles it... going to learn Nuke. Seems like it's the way to go.

And agreed on 4k vfx. I'm glad I'm not the only who feels that way. The render times and ram requirements add up. That said, those lenses aren't ever in my budget and the distorted look of anamorphics is what appeals to me visually if not technically.
Click to expand...
Nuke is great, but real expensive ... the recent caravr add on was over $4k just for a simple plugin with stuff that comes free with a bunch of the new graphics cards. Also with nuke you really need to use nukex, since a lot of the lens functions are only in nukex (which is a bunch more change). The premiere/AE combo I think can do anamorphics real well (the smashing/unsmashing is fairly seemless). If I was a heavy davinci user I'd also check out fusion. Both nuke and fusion have free "learning" editions. Expect around a year to really get efficient with nukex.
 
You must log in or register to reply here.
Back
Top