Andrew Somers
Forum Replies Created
-
There are many reasons NOT to shoot at 60fps all the time when you are finishing in 24fps;
1) If you want to playback 60p footage at 24p at “normal”speed, then frames have to be removed, and it is not a 1:1 relationship, so at 24fps some frames are temporally closer together than others, resulting in a juddering motion which you normally do NOT want.
2) Shooting everything at 60 results in a much greater amount of data and storage needed.
3) Your shutter duration at 60 will typically be different than what you want or is optimal for 24.
4) 60 typically requires more light for a given EV due to the shorter shutter duration.
5) Depending on the camera system/type and the type of file you are recording out to, there may be other problems shooting at the higher rate, such as increased compression artifacts.
IN SHORT: especially if your target is 24, shoot everything at 24 that you want to play at 24, and only shoot higher frame rates when you want a slowed-motion effect. This is the kind of thing you should be in touch with before you shoot as a matter of planning.
ALTERNATIVES AND EXCEPTIONS: There are flow field post processes (i.e. “timeWar”, :”Kronos”) that can create a credible slow motion from normal speed elements (depending on the subject matter). These processes CAN also be used to get a better interpolation to make 60 play normally at 24. However, they are not without artifacts and do take a lot of rendering time to complete.
If your target is 30, and your story has a lot of speed ramps/cuts back and forth from normal to slomo, then you might be able to make a case for shooting 60 for all shots for scene where that is needed.
If the target is 24, the equivalent would be shooting in 48, or maybe 72, though I’d expect greater problems in post, including and not limited to audio sync issues.
Andrew Somers
VFX & Title Supervisor
https://GeneralTitles.com -
Hi Gus, here ae a couple ideas:
1) Get and use the Real Smart Motion Blur plug in – it allows for creating vector maps to control the amount of blur in a particular area, among other things, but can also over-accentuate the blur of a moving object.
2) Try a combination of directional blur on an adjustment layer with feathered masks to hold out areas you want less or no blur on.
3) Use the built-in motion blur as I describe in THIS ARTICLE on how to use AE to create traditional optical printer aerial imaging effects.
Andrew Somers
VFX & Title Supervisor
https://GeneralTitles.com -
As a matter of practice, I recommend importing PS files and placing into a layer at the native resolution of the PS file. Then place that entire layer into another comp that is at the output resolution, and scaling the PS layer as needed.
Andrew Somers
VFX & Title Supervisor
https://GeneralTitles.com -
Hi Danielle,
This is a common issue, it is due to some combination of: inconsistent color profiles being used, uncalibrated monitors, Not using color management, and even color profiles with incorrect gamma curves.
Here’s an article I wrote that covers this subject. And if you could tell us what all of your color and project settings are, we could better help.
Cheers!
Andrew Somers
VFX & Title Supervisor
https://GeneralTitles.com -
There are several cloud rendering services you can use that support After Effects:
https://www.pixelplow.net/support/
Not to mention that you can get AWS servers configured into as many render nodes as you care to pay for using E2 instances.
Andrew Somers
VFX & Title Supervisor
https://GeneralTitles.com -
Yea I agree there are plenty of times when straight is better than premult. In fact I was so used to working in straight that when EXR came along I remember being surprised that they went that way. Nevertheless, I see their rationale as to the value when working in scene linear.
I’ll pause briefly to mention that After Effects (and PS) don’t handle alpha channels in linear light “quite right” — AE has some definite issues with opacity not functioning as expected when converting to or from a gamma encoded environment for instance.
Nevertheless I use the UNMULT plugin to convert to straight alpha in AE — something so basic and simple I think AE should just “do it”.
Andrew Somers
VFX & Title Supervisor
https://GeneralTitles.com -
Hi Steve, have you modified your Interpretation Rules?
The rules has this line near the bottom:
# OpenEXR is premultiplied by convention
*, *, *, "oEXR", * ~ *, *, *, P, *, *
You can find it here:
~UserName/Library/Preferences/Adobe/After Effects/11.0/Interpretation Rules.txt
Just FYI, the EXR spec defines the RGB channels as premultiplied, it’s actually part of best practices/recommendations for use. Here’s what they have to say on the subject, starting on page 20:
Nevertheless, if you want it to always be STRAIGHT, then you can make the line read:
# OpenEXR set default to interpret alpha to straight.
*, *, *, "oEXR", * ~ *, *, *, S, *, *
I set that as a SOFT rule (using the tilde ~ ) so if the file has metadata that specifies premult, then it will come in premult.
In Interpretation rules you can also add in the color profile, frame rate, and linearization to what you most often use. Here I set it as a hard rule (using the equals = ) so this line overides embedded meta data.
# OpenEXR set default to interpret alpha to straight, 2398, sRGB Primaries and Linear.
*, *, 23.976, "oEXR", * = *, *, *, S, "sRGB", 1
Hope that helps.
Andy
Andrew Somers
VFX & Title Supervisor
https://GeneralTitles.com -
The first thing I recommend is to set your project to 32 bit and linearize the colorspace. When working in linear mode, you can make things brighter than 100% white, which would clip in a gamma encoded space like Rec709. In linear those overbrights and their detail are maintained — if you bring them back down with another Levels effect, you’ll see the detail is not lost.
Blending light in scene-linear space works like it does in the real world. Instead of adjusting opacity, you can use ADD transfer mode to add layers together, the light values of each layers adding together like real light does. You’ll find using ADD to place the glowing smoke onto an image gets you toward your goal.
The LEVELS effect will act a little differently than you’re used to — you can make things brighter than white as you ADD layers together and then in an adjustment layer another Levels to bring everything back down and still retain detail.
Also, if you make the smoke a dark blue, when you use Levels to make it brighter, it will end up glowing white as it gets bright. Put a Gaussian Blur on it to smooth it into a plasma look. Adjust the gamma of the individual channels with Levels or Curves to get some interesting color shifts between dark and bright.
A possible layer setup might be:
Layer 1: Adjustment layer with LEVELS to adjust the overall output.
Layer 2: Particular layer. After the Particular plug-in, put the LEVELS effect then the BLUR effect. Set this layer to ADD transfer mode. Adjusting the gamma of the individual RGB channels, as well as their in/out levels can create some interesting color shifts based on the brightness of each channel. You can take this idea even farther using the CURVES effect.
Layer 3: This is the footage you are comping the smoke on to. Make sure it’s color profile is set/interpreted properly so that AE color management converts it to the linearized working space. You may want to have a Levels control here too.
Remember that in linear space, especially when using ADD, you have a much greater range of adjustment than a gamma encoded space. You might spend time getting the two lower layers comped together nicely only to realize they are overall too bright — never fear, you can leave those adjustments alone, and just use an adjustment layer to bring everything back down.
Andrew Somers
VFX & Title Supervisor
https://GeneralTitles.com -
Hi Shahed,
The first thing I’ll tell you is that the first of the two above (wwgaYb.jpg) is tagged with an icc profile called “linear color space”, implying a gamma of 1.0. The second file is NOT tagged at all, and untagged images are generally interpreted as sRGB which has a gamma curve close to 2.22
Photoshop is by default working in a color managed workflow, interpreting color profiles embedded in the image to convert them to the working space (unless you turn this off).
After Effects however requires you to turn on color management by setting the working space to a profile.
Based on these images, I’m going to guess the settings that would be best for you:
In After Effects, set the project to 32 bit, sRGB, and Linearize Colorspace. Also select “adjust for scene referred profiles“. Under “View” set “Display Color Management” and finally, set the interpretation for the image sequence you are importing to linear (if it is linear or convert it to linear if it is gamma encoded) in the clip’s color management tab (o using the convert profiles utility effect on the layer).
It’s common to render in a scene-linear light (gamma 1.0) in C4D, as scene linear is the way light works in the real world. And it’s best to remain in a linearized working colorspace all through the compositing process until your final output.
NOW,
Having said all that, you do not want your C4D linear renders to be saved as linear 8 bit !!! That would be bad and have bad banding types of artifacts as 8 bit cannot hold the HDR nature of a scene-linear image. Ideally you want to save them to a file format that is natively linear and at least 16 bit. And PSD is not the ideal way.
But EXR is.
C4D outputs to EXR, and AE supports EXR files with the OpenEXR plugin and the ProEXR plugins. I believe these Fnord plugins are all included for free in AE. But you can also download them free from Fnord, along with documentation on using them.
The ProEXR plugins include one called EXtractor which allows After Effects to access the UNLIMITED number of channels that EXR files support (much like PS Layers).
I recommend 16 bit float EXR as the preferred interchange format for VFX and CGI. Working with C4D it’s ideal when you render in a linear light space (gamma 1.0 — I believe it’s the default for C4D), so you can export in linear light which is the native transfer curve for EXR. Also EXR supports channels of different bit depths in the same file — 16 bit float is enough for images, but you might want a 32 bit channel for Z-Depth for instance.
As the native/preferred gamma of EXR is 1.0, you can import direct into AE 32 bit linear mode with no transfer curve issues — nevertheless it IS still important to know the color primaries that were used rendering in C4D. Since it sounds like you might be doing things at default values, the primaries are probably Rec709/sRGB (these two profiles use the same primaries).
As for your current workflow, In AE you can change the way the imported footage is interpreted and assign it any profile you want. Right click the footage and select “interpret footage”, then set the color management settings with the appropriate profile. Or you can add the utility plugin “Convert Color Profile:” which allows you to set input/output profiles and control linearization.
*** If your AE working space is scene linear with Rec709 primaries, and your footage is scene linear EXR with Rec709 primaries (essentially the EXR default), then you can set the imported footage to “Preserve RGB” which turns off color management for that clip as it’s not needed, improving performance by avoiding color transforms.
A further tool for color, LUTs, and profiles is the free OpenColorIO, downloadable here.
And finally, I wrote this article discussing working vs output color spaces and linear workflows that you might find helpful.
Best of luck
Andrew Somers
VFX & Title Supervisor
https://GeneralTitles.com -
Hello Horst,
A couple things:
1) The majority of quality is lost on the FIRST iteration of a DCT type of compression. After that, even if it is re-compressed, the quality loss is minimal (assuming the same data rate.)
2) Sorenson is a dead and discontinued product. Adobe Media Encoder (AME) is just as good and integrated into the Adobe workflow. AME and Sorenson BOTH use the MainConcept codecs for mp4s if I am not mistaken, but I think you’ll find AME is better/easier/faster. If you have a reasonably fast computer with multiple cores, expect AME to fly through the render at close to real-time speeds (if not faster), depending on the frame size, frame rate, etc.
3) How long is this video? You’re making it sound like it’s a huge hours long piece! If you have the lossless file, I’d suggest re-syncing the new audio in AE or Premiere, and outputting using AME.
4) If you do the quicktime trick, it is CRITICAL that you SAVE AS and make certain to select “make movie self contained”. If you don’t, the QT will not play unless the source files are with it.
5) I’d be more inclined to swap the audio on the lossless version, then compress that in AME.
Related comments, but not what you’re asking about:
A) The “lossless” codec is a joke, and it’s not worth using. It’s 8bit, and you’ll get better fidelity using a 10 bit ProRes codec as your “master” render output. Other things being equal, bit depth is moe important that compressed vs uncompressed. And the prores file will be smaller, so it will be easier to handle and will render faster due to the lower dats bandwidth.The “lossless” codec is a data hog, and you don’t get any real benefit for that data space it eats up.
B) I rarely use the audio features in After Effects. It is so much better to work with audio in FCP, Premiere, or Avid. My suggestion is to put the lossless video into Premiere, sync the new audio there, and then send it from Premiere to Adobe Media Encoder for output to mp4.
C) Since it sounds ike quality is important to you, in AME, make sure you set “two pass encoding” and set as high of a profile as you can, relative to the final file size you need.
Best of luck!
Andy
Andrew Somers
VFX & Title Supervisor
https://GeneralTitles.com