[Anthony Sipsas]

Never thought of that, ill give it a try thanks!

Anthony Sipsas
Motionographer
Los Angeles, CA

[Anthony Sipsas]

The only conclusion I was able to came up with was that the Sony F800 we were using to film (at a smaller shop that had older equipment) was not very effective at properly achieving pulldown. There were several settings on the F800, and the DP happened to want to get the footage @ 23.976. I have no idea why, as we were never consulted in post to what frame rate we would prefer to work in but the DP chose 23.976, and thats all i had to work with. we did not have the budget or time to reshoot. The main thing to know in that situation though, is only after researching the F800, we found that the camera only films in 29.97, so any other frame rate/setting is going to have problems as its not native to the internal workings of the camera.

As far as the physical process of getting 23.976 footage out of the SONY F800, as I said it actually films in 29.97, does a 3:2 Pulldown (as Dave had suggested earlier), within the camera before it records it to XDCAM. You cannot override this process.. If you are on an SONY F800, it natively records in 29.97, so it would probably be best to save to disc @ that framerate as to not add in more “problems” into the pipeline. I have always personally believed that one should always work in the native frame rate of the camera all the way through post, and only change frame rate for delivery as to reduce any variables that may complicate things. I am not sure how this would play out for something thats timed to music (such as a film though), because most of my work has been short form (Interactive, web commercials).

I would still have to stick with what I said on not changing FR as I personally believe that its the DPs responsibility to speak with a post supervisor, and/or research whats going to happen to be needed to be done technically down to delivery, and choose to make a decision with their camera that can balance art vs technical necessity/practicality.

The “Samurai” from atomos was also completely useless. Both (XDCAM and Samurai) are set to 1920×1080 23.976 in camera, and reflect that when i check the info on the clips.

The problem is that both the camera and the Atmos labeled 23.976 as progressive when its not. 24p is progressive, 23.976 is not. This is why i question why would a camera like SONYF800, or an ATMOS product label it as such? They aren’t fly by night companies by any means…

Again, reading my previous posts, it sounds funny, but I can assure you I was objectively reading the settings from the camera. This is not a reflection on my knowledge, but rather a reflection of what I am reading from the users manuals, and in camera settings.

As far as your DSLR, I would do a few things:

1) Research if your DSLR films at a specific Frame Rate. What Id like to eliminate here is when you switch to a different frame rate, if the camera is natively filming, and writing to disc/card @ that frame rate, or if there is a software process thats being added on top of a specific format that the camera can do. What I mean by that is if for example the camera can only film @ 24p, (and this is stated by the manufacturer as the sole capability of the camera natively), that would mean when you switch it to 60fps, 30fps, etc that there is some sort of processing going on within the camera that alters/converts the footage to the other setting before it saves it to the card.

Obviously there are reasons to do one over the other, but its probably good to be aware of this, as it can affect your footage. Its probably smart to know this information about your camera so that you can choose its native settings and change anything after post (when you deliver), unless something is really needed like going from 24p- 60p in camera for slow motion in post or similar example…. But then you know that you are choosing to do something that may affect the footage, but its necessary. The idea is to avoid problems if they are unnecessary.

2) The other issue was the F800 had a LOT of noise in the footage. Make sure your camera settings are such that you are receiving proper light, and proper film speed. I don’t believe the Cannon cameras have these settings adjustable, so probably just make sure your scene is properly lit. I know cannons have amazing low light sensitivity for DSLRs, but in comparison to pro cameras, they are still very noisy.

3) The last thing I would do, is I would add an extra step in your workflow specific for tracking. What I mean by this, is SPECIFICALLY for tracking, I would export an uncompressed frame sequence at the same framerate as the native footage. Import & Use the image sequence for tracking, but only use it to acquire the tracking data. Once you have your tracker solid and sticking, save/bake, & replace with source footage, and finish compositing as needed.

**we can only do the best we can with whats given to us, so hopefully the last step will help you deal with the footage you have already shot. Keep the other steps in mind (food for thought) for knowledge so that you can make decisions that will make your workflow more fluid moving forward.

Hope that helps a bit. Please keep me posted on your findings as well….

[Anthony Sipsas]

Footage is coming out of the camera @ 23.976fps.

I am researching the XDCAM information and here is what I came up with as far as what the cameras pdfs are telling me:

PDW-F800 (SONY)> XDCAM:

Mode (Codec): MPEG HD (MPEG-2 MP@HL)
Pixels: 1440 x 1080
Y/C Sample: 4:2:0
Frame Frequency: 23.98p

All while there is a Samurai recording same signal to flash drives in real time to:

Compression (codec): Apple Pro Res
Pixels: 1920×1080
y/C Sample: 4:2:2
Frame Frequency: 23.98p

I bring both clips into AE, and they both read exactly what I Typed here. (XDAM comes in @ 1440, while Samurai comes in @ 19020×1080)

My question is, is the samurai doing anything to it funny maybe?

I havent tried a track on the XDCAM footage straight from the camera yet, but I watched the footage on the XDCAM taken straight from the sony media, and it doesn’t look normal. Motion blur on objects moving fast are solid (no transparency). I’ve never seen that before.

The footage on the solid state looks fine going through it frame by frame, but I am having the issues with the track that Im talking about above….

Ill be doing more tests throughout the day.

Thanks again.

[Anthony Sipsas]

Thanks Guys for all your help. Track Fields is not enabled, so I dont believe thats doing it.

As far as the conversion, the information I found was from quick internet research, so it may or may not be accurate, but Im leaning toward what Dave is suggesting about the 3-2 Pulldown. Unfortunately, I am looking at the footage straight from the camera, so anything that I am experiencing is done in camera as Im pulling it straight of the XDCAM media.

There is also a “Samurai” from atomos (Similar to this:
https://www.bhphotovideo.com/c/product/858661-REG/ATOMOS_ATOMNJA002_Ninja_2.html
)

Both (XDCAM and Samurai) are set to 1920×1080 23.976 progressive in camera, and reflect that when i check the info on the clips.

Reading the manual (Large and useless) from the Sony F800, it stated something about certain frame rates at certain settings being artificially generated, and not true settings (which is a tendency in some cameras).

Ill take a look at it again throughout the day today, and see what I can find.

IN the meantime, I see a setting I have never used when interpreting footage (within AE) that states “24pa PUlldown”. What is this as opposed to 3-2?

Thanks again guys, and thanks Dave!

[Anthony Sipsas]

Never heard that term before. What does it mean ? Oh wait, do you mean fields as in interlaced footage? Ah ok, no this is Progressive. Where can I find this option to disable it?

[Anthony Sipsas]

Dave,

I also did some asking around to find out that the footage is XDCAM. Im not familiar with this, but with a little looking around, I read that XDCAM footage is really 29.97 being “pulled down” to 23.976 in camera? Is this true? This would explain what I am seeing, and what you were describing, but im curious if this is the camera doing this? If so, would I need to interpret for pull down?

Camera:

PDWF800 Filmed @ 23.976p 1920×1080

Wow this is an odd situation….

Thanks in advance,

A

Also, to show a little bit more on “inconsistencies”. Take a look on my timeline (further down from before), as you can see, after a certain point, AE is no longer key framing data on frames, but rather solely on in between frames, and is doing it inconsistently. You can also see how it also doesn’t key frame ALL the data:

[Anthony Sipsas]

Dave,

So i re-imported the 23.976 footage, into a new 23.976 pre-comp and tracked.

Im still getting half frames. Its inconsistant too. Sometimes it does it, sometimes it doesn’t.

Heres the info”

If you look closely at the keyframes outside of the close up area I have high lit, you can see that there is no mid-frame data in that area. Which shows an inconsistency that I wanted to point out.

The other thing is, when adding the “mid-frame’ keyframes, sometimes it adds all data, sometimes it only sets keyframes (on a mid frame) for feature center, or Attach Point, but not Confidence or All 3 (as it does on the “On Frame” Keyframe data, which leads me to believe that its not an intended feature.

Let me know what you think.

[Anthony Sipsas]

My only guess would be to pre comp in a 23.976 pre comp just for the track, then put that pre comp Ina 24 timeline?

[Anthony Sipsas]

Thanks Dave. Clip is 23.976fps, -not 29.97. and am aware over long periods of time , sound will loose sync on 24fps timeline.

I have not dealt with 3-2 pull down much in my career being that I work mostly in interactive space and for web, and clips given to me are usually 24p or 60p.

From what I understand, 3-2 is when you try to play a 29.97 clip on a 24fps timeline? Is that correct?

I was getting mid frames on the 23.976 prior to interpreting to 24. Thought interpreting to 24 would solve why I was getting mid frames key frames on a 23.976 clip.

Same results, so I reverted to 23.976, but still have a mod frame data issue.

Any ideas?

[Anthony Sipsas]

Thanks, Guys, i kept messing with it, but didn’t get it to work until i did the following:

x=(ease(time,wiggle(.5,2),wiggle(.5,2)));

[x[0],x[0]]

I guess i was missing the “[0]”

is there a “dictionary” so to speak for expressions in AE so I can at least make sure my syntax is written in a means that AE can understand?