PDA

View Full Version : 24fps vs 30fps



inquisitive
10-13-2006, 03:23 AM
Are most animations rendered at 24fps? What do you guys use?

Ztreem
10-13-2006, 03:39 AM
I'm not sure but I think movies are at 24 fps and PAL is 25 fps and NTSC is 30fps.

Auger
10-13-2006, 08:40 AM
24fps. Looks much more filmic (is that a word?) and of course renders faster.

I only use 30 fps if I need a real video look. Maybe a logo or on-screen display sort of thing. Or, if you need to match into pre-existing video footage.

Jon

Sekhar
10-13-2006, 10:32 AM
Yes, depends on the footage with which you'll be mixing the LW output. 24p (progressive) is pretty much the standard in video these days (even 30i footage usually gets de-interlaced to 24p), so 24fps is the way to go in most cases.

If your footage is 30i (interlaced), of course you'll render to 30fps, but also consider making it interlaced to maintain the look - if you do that, be sure to match the field order (lower/upper). Progressive and interlaced have a different look that can't be fully fixed in AE, etc.

TheDynamo
10-13-2006, 01:00 PM
I would be nice to keep a completely digital flow but once that darn 24p reaches videotape or the avid (I'm not quite sure where the originator is at my workplace) you have to deal with the rasafraggin fields. Granted AE does handle the problem with tweaking settings, it just opens a few more possibilities for bad renders.

-Dyn

(thinking of ways to combat the use of interlaced video without the use of violence) :D

bluerider
10-13-2006, 04:46 PM
Are most animations rendered at 24fps? What do you guys use?

Classical animation is based on 24 fps. All the feature animated films you see are using this system for their animators.

Its certainly an easier number to divide up. When it comes to output for region NTSC, PAL or film etc you just convert the frame rate to match that medium, a post production process.

ScottSullivan
10-13-2006, 09:19 PM
I missed the obvious answer already mentioned a few times, but yes, 24 here, too. Those extra 6 frames per second you save by not having to render them add up fast!

Even when dubbing down to miniDV, the rendertime at the Print to Tape phase is MUCH faster than those 6 fps, especially on bigger scenes.

Scott

KevinL
10-13-2006, 10:13 PM
2 points.

1-In NTSC land, std def 30i is still the standard and a lot of people do not have 24p infrastructure through and through.

2-24p has temporal artifacting, relative to 60fps (30i's real frame rate, albeit at 1/2 rez)

Have fun....

kief
10-15-2006, 02:04 AM
2 points.

1-In NTSC land, std def 30i is still the standard and a lot of people do not have 24p infrastructure through and through.

2-24p has temporal artifacting, relative to 60fps (30i's real frame rate, albeit at 1/2 rez)

Have fun....

I am guessing the temporal artifacting you are speaking of is caused by the added pulldown (ie 2:3:2:3 or 2:3:3:2) which does not exist in actual 24p if i am not mistaken. it should only be added if it is to be viewed on an NTSC TV or monitor.

For those who may not know. A pulldown or pulldown pattern is basically a way of duplicating certain frames or fields of frames so that true 24p can be played on a NTSC device which, as stated earily, plays 30i.

loki74
10-15-2006, 03:26 PM
Temporal aliasing and artifaction should not occur in any footage shot at or animation rendered at 24p native.

As a rule of thumb, when producing for TV and I want it to look like TV, I use 30. If I want the film look, I use 24. Ultimately though, it will be best if you match the fps and (if applicable) field order of your other assets, especially if you're planning on doing match moving or compositing.

Re: pulldown: Just to clarify, I'm pretty certain that 2:3 is for making it look smooth (ie putting your 24p footage onto a 30p DVD) while 2:3:3:2 is for easy lossless coversion back to 24 (ie filming at 24p with an NTSC device to 30p and converting it back to 24p for editing and compositing on your computer or printing to film). ...correct me if I'm wrong

kief
10-15-2006, 07:54 PM
loki74,

re: pulldown
that's exactly as i understand it.

KevinL
10-16-2006, 12:43 AM
By temporal artifact, I am referring to the fact that you have a "lower" sample rate in the time domain for motion delivery. While not 100% the same, there is a well understood phenomenon in audio sampling called the nyquist frequency, where the workable frequency is half the sample rate. Frequencies over the working resolution limit that are allowed into the sample will cause frequency aliasing.

Forget for the moment Lightwave, digital animation, rotoscope, film, video, progressive, interlaced... etc.

I speak strictly of the time resolution.
It’s late, I’ve been working a long day so I ramble a bit..... but maybe this will give a useful hint: two examples
1-Films with the stage coach wheel spokes that “freeze or go in reverse” because of the 24 fps sample rate interacting with the frequency of the wheel rotation (look at the gate on many projectors, 2 windows on the disc= perceived rate 48 fps this reduces flicker, but doesn’t change the motion sample rate)

2-Overall resolution= Height x Width x Frame rate, so simplistically speaking 24p has lower overall resolution (in time domain) than 30i (which actually is 60 fps, with each “frame” having half the vertical resolution) .

Now before you think I’m knocking 24fps, I’m not. I use it often, sometimes I do it on the twos (12fps) and other times have to render at 30fps. The issue of legacy frame rate conversion, delivery modeling (ie. interlace to progressive or progressive to interlace) and all of the other fun stuff is all part of the overall consideration for a project. Just now wrapping on a 30fps project to match move a plate, while two weeks ago I rendered an 8fps for an oddball delivery format. and next is the live action 30i, I have to CG and MM, then we go out to 30i, 24p and who knows what the **** else.
Have a moving experience :)

ps. A usefull exercise for the whole interlace/progressive, different frame rates thing. Look at a film thats out to 30i on DVD. As you step and see the stutter frames (repeats to make up data) some are whole original frames and some are two frames sharing each of a field for the final output. That is artifacting (false data or not true to the original). A fun posit: real life might have a frequency (how high? ...infinity? ... hmmmm) and everything else "samples" that. Eyeballs, smell, a photosonics camera, varicams, we humans emulating life when we animate, and all the other stuff......

bedtime!

DiedonD
10-16-2006, 02:36 AM
Temporal aliasing and artifaction should not occur in any footage shot at or animation rendered at 24p native.

As a rule of thumb, when producing for TV and I want it to look like TV, I use 30. If I want the film look, I use 24. Ultimately though, it will be best if you match the fps and (if applicable) field order of your other assets, especially if you're planning on doing match moving or compositing.


See this is the part that it gets confusing for me.

If its for TV its 30, if its for movie its 24. But what if its about a movie on a TV? The movie will be shown on a TV, then how many frames do you go by?

Its the same about DVD's. They too have a certain fps, and the TV has 30fps. But what if you'll be playing the DVD on a TV?

inquisitive
10-16-2006, 04:44 AM
Thank you all for the responses.

I have not gotten around to render anything serious, but while visiting an animation studio they talked about rendering at 24fps, and up to that point i was thinking that 30fps would be the way to go for my project.

I may mix dv and animation or I may just go all animation, still working on the story. It's just a personal project that if I ever get to complete (as you know it all takes time) I would like to submit to some short film contest.

I bumped into some article after I posted this question (can't remember the URL) that suggested to render at 24fps, and also render in widescreen format, instead of full screen to save even some more render time.

starbase1
10-16-2006, 05:22 AM
OK, I read kevinl's post and I am none the wiser...

I think it is a slightly different question, but seems definitely relarted to me.

How do they go about coping with the differeng end frame rates when they do those all CGI films, Toy Story et al?

I assume that they don't render the entire thing at multiple frame rates, so how do they switch between them? Of course the studios are not using 50 quid DVD authoring software, but mine just drops or repeats one frame in six, which makes the whole thing stutter slightly. And I cant think of a better way of doing it.

Nick

KevinL
10-17-2006, 09:45 PM
Sorry for not contributing to your wisdom :)

If you are animating for film (principle output) then 24fps. The video on DVD is either treated as:
A-Convert to interlaced 30 (more control of the interlace recipe)
B-Print as 24fps and leave it to the DVD player to do the 24p output or up convert to 30i

Hope that one's clear
Kevin

MiniFireDragon
10-18-2006, 11:15 AM
A clarifing point to NTSC. The number is not 30 fps. That number is actually a rounded number. True NTSC spec is 29.97 interlaced.

Intuition
10-18-2006, 11:29 AM
I wanted to chime in since I do animation for both types.

A few points.

1. 24 FPS is only going to render faster IF you render at NTSC resolutions.

Typically you would render a film look (24fps) at a film resolutoin like 1920x1080 or even the native film which is somewhere between 2k to 8k depending on the film resolution that it is scanned in to digital.

So if you are rendering 24fps and plan to play that on television you have to do a 3:2 pulldown conversion which converts the 24fps to the 30fps (29.97ntsc). It will have the strobe effect but the viewing area will be not as wide as true film or High def.

2. High def native. or Film scanned resolutions

In High definition you do render out 24 fps which is less then 30 FPS but you then have to consider that a 1920x1080 image is almost 6 times as much informatoin as 720x486 ntsc. SO you render less frames but each frame would typically take about 6 times as long to render (average) as a 720x486 frame.

3. If your only playing the animation on quicktime then just render 24fps. Only consider the above if your format is going to play on TV or in a film.

Remember...

24fps is typically for film = less frames but higher resolution making a 24 vs 30 frames a null point. The strobe and 16x9 frame will look more filmic and higher quality though.

30 fps is more frames but less resolution.

You can however keep the 16x9 and 24fps but use the NTSC. This is called letterbox. Letterbox is confusing to some people. They think that the black lines on the top and bottom are stupid. Yet they don't realize that the wider format of film, if being made to show the entire width of content on the TV viewing ratio (4x3), would either have to have the sides off screen (called pan and scan) OR have the full frame which would show black on the top and bottom.

If you took a film frame and tried to scale it so that the width is fully in the screen but allowed the top and bottom to fill your TV it would make the image look squished.

These are standards for broadcast or film. But, if the animation is not going to end up on these formats then 24fps at any resolution is fine for viewing.

starbase1
10-18-2006, 01:45 PM
Is there a reasonably priced bit of software that will convert inteligently, rather than just duplicating or dropping frames?

Sekhar
10-18-2006, 02:58 PM
Not sure what price range you had in mind, but Magic Bullet (http://www.redgiantsoftware.com/magbulsuit.html) does a good job of de-interlacing (e.g., 30i to 24p) and is very popular. AE can do it natively, but the results aren't acceptable in most cases.

ColinSmith
10-18-2006, 04:36 PM
This is more of a 1 trick program than the Magic Bullet plugin but supposidly give a bit higher quality deinterlace and renders faster and won't need a compostiing program to run in.
FilmMaker (http://www.dvfilm.com/maker/index.htm)

To get a decent 30fps to 24fps conversion though you will need it to be 60i, 30p will not do much good.
This software can also apply 3:2 pulldown to make decent 60i from 24p footage if you go in the reverse direction.

They also do a good PAL<->NTSC program that works a similar way.

LFGabel
10-18-2006, 10:37 PM
Thought I'd post this to help explain the telecine process, which is converting 24p to 30i...
http://en.wikipedia.org/wiki/Telecine#Common_pulldown_patterns

loki74
10-18-2006, 11:05 PM
See this is the part that it gets confusing for me.

If its for TV its 30, if its for movie its 24. But what if its about a movie on a TV? The movie will be shown on a TV, then how many frames do you go by?

Its the same about DVD's. They too have a certain fps, and the TV has 30fps. But what if you'll be playing the DVD on a TV?

Anything on any NTSC device will NOT be true 24p. If it was a film put on DVD, chances are it was resampled at 2:3. Hence, if you want to make it look like film put on DVD, render, edit, composite, etc at 24 and resample your final footage to 30p with a 2:3 pulldown ratio. If you want to make it look like TV on DVD, go 30p right off the bat.

Re: temporal aliasing

Ah, okay, now I know what youre talking about... if we have a revolving wheel which makes one full revolution in 1/25 of a second, sampled at 24p it will appear to be moving slowly backward. This is true, but it seems to me that this is a well-recognized effect, that is, the audience will know that the wheel is spinning forward, but looks like its spinning backward because thats "just what happens when stuff moves fast." People are familiar with it I think, and I feel it contributes to the ever-passionately-pursued "film look."

Now generally, lower resolution is associated with lower quality. But consider a graph comparing quality to resolution. In the spatial domain, I would say that resolution increases quality pretty much linearly. In the temporal domain however, I would say it increases logrithmically. Another way of looking at this is to consider a limit at which differences in resolution become inconcieveable to most audiences. I would argue that this limit is reached much sooner in the temporal domain than in the spatial domain. A practical example: 480 vs 4K = drastic difference. 60K vs 60.6K = not so drastic. 6p vs 15p = drastic difference. 24p vs 30p = not so drastic. The point is, obviously 480 and 4K are VASTLY distant from 60 and 60.6K, (ha, like we'll EVER see 60K in our lifetimes!) while 6 and 15 are not all that far off from 24 and 30.

starbase1
10-19-2006, 03:06 AM
Thanks again to all, I'm still getting my head around this!

MiniFireDragon
10-19-2006, 10:39 AM
Programs to de-interlace, if that is what you were speaking of, use Virtual Dub (Free) or IUVcr (~$40) and the Alparysoft De-interlacer (which comes built into iuvcr).

Virtual Dub: http://www.virtualdub.org/
IULabs: http://www.iulabs.com/
Alparysoft: http://www.alparysoft.com/

KevinL
10-19-2006, 06:58 PM
I have known a fair amount of the information contained in this thread for a while (as much as needed) but by no means all. I am finding all the input and discussion to be excellent. This is my ideal for any thread I read or participate in. I had pretty much retired to lurker because of the negative experiences in various groups. This gives me hope :)

Thank You,
Kevin