Retiming & transcoding from 29.97p to 25p

Hi, dear Community!
I have an issue which I'm failing to solve:
I have a film in NTSC SD 29.97p which I need to transcode to PAL SD 25p.
But what happens:
1) When I place the film on the timeline of my FCproX in its native 29.97p I see repeated (coupled) frames every 4 frames
2) When I place the film in 25p I see coupled frames in order 3+3+3+12 frames
3) The only way up to now when I see a frame by frame sequence when I place the source video to 23.98 and apply Nearest Neighbour mode from Rate Conform pane.
But very interesting happens further:
When I bake (render) a file from 23.98 Optimised and then import the new file to the timeline of 25p it automatically compresses from 01:30:55:00 to 01:27:15:20 (it equals to 4,17% squeeze) and I can do nothing with this - when I apply SloMo to compensate this shortness I see the coupled frames again.
Maybe something wrong is with some metadata inside the file with the film?
I appreciate your help.
Best regards
Ol

Use Compressor.

Hi Ben!
Many thanks for the advice.
I tried Compressor to extend the file and made
BtoD_720x404_25p_10bit noncompresed_1.00x_720x404_alq-12-Apple ProRes 422 HQ.mov
to conform the initial timing, but unfortunately still have coupled frames, every 25th frame

Hi Ben!
Many thanks, but this option doesn't work in this case.
Probably something is around interlacing or pulldown.

I offer a bit of clarification on frame-rate conversion, as this seems to come up regularly.

When you place a 24fps or 23.98fps clip in a 25fps timeline, FCP places each frame to coincide with the 25fps rate. Therefore, no frame interpolation, blending or other picture turmoil is needed, and, yes, the clips will run faster by the difference in frame-rate and the audio will increase in pitch. This is the easiest way to handle these close frame rate differences, so it's done this way.

Similarly, placing a 25fps clip in a 24 or 23.98fps timeline will have the opposite effect, i.e. the clip will slow down and the pitch of the audio will be lower. You can, of course, use the Pitch Change audio plugin in order to correct for these differences, if necessary.

However, when trying to accommodate 29 or 30fps clips in a 25fps timeline, as in your case (or indeed the opposite), you generally only have the choices of horrid, bad, not too bad or 'if that's the best it will do'.

Years ago, this sort of conversion was done with an expensive bit of hardware called a Standards Converter, and some makes and models did a very good job of disguising the frame rate conversion problems, well, mostly.

Now that Apple Silicon is here, maybe Apple could approach frame-rate conversion a bit more head-on by using the new computing power available?

It's a perennial problem.

The duplicate frames are necessary because you'd have blank frames otherwise. Something has to fill that space. But it shouldn't be that noticeable.  I'd like to see side-by-side samples of what you're doing.

FCP should not be changing audio pitch at all, technically. If you find it is, that's a bug, as Apple claims no pitch changes.

OlRight wrote:

I have a film in NTSC SD 29.97p which I need to transcode to PAL SD 25p...When I place the film on the timeline of my FCproX in its native 29.97p I see repeated (coupled) frames every 4 frames...When I place the film in 25p I see coupled frames in order 3+3+3+12 frames...

The fact that alleged "29.97p" material in a 29.97 timeline shows every 4th frame as repeated implies it's not true 29.97 material -- or didn't start that way. Actual 29.97 material in a 29.97 timeline should be smooth, with an even cadence.

It's possible someone started with a 24 fps clip, imported to a 29.97p timeline which by default will duplicate every 4th frame, then they exported that, baking in that to the output file. Or else it was a capture from a TV source, e.g, a DVR which had already done rate conversion from 24 fps to 29.97.

Then you imported that 29.97p clip to a 25p timeline and are trying to make it smooth. You can try optical flow rate conforming which might help, but the problem was caused by whoever baked in the repeated frames to the 29.97p clip.

True 29.97p material will usually work somewhat well in a 25p timeline -- if using optical flow rate conforming.

24p and 25p material can be smoothly rate conformed with a rate change, which is what FCP automatically applies in that case.

It appears the problem is 24p material was exported as 29.97, which locked in that frame cadence, making it difficult to fix. This is similar to cases you see on Youtube where someone exported an interlaced file as progressive, thus baking in interlacing artifacts, making deinterlacing very difficult or impossible.

Hi Joema!
I've understood it intuitively and having done so many tests already:

"It appears the problem is 24p material was exported as 29.97, which locked in that frame cadence, making it difficult to fix. This is similar to cases you see on Youtube where someone exported an interlaced file as progressive, thus baking in interlacing artifacts, making deinterlacing very difficult or impossible."

I tried Reverse Telecine in Compressor, but unfortunately it doesn't work as I recognised that someone before sending the file to me cut the beginning (I do not see the very beginning of the film, a studio caption)

I tried Amazon online MediaConverter, they have so many options to choose  and I already tried about 10 different combinations of codecs/containers/filters and special filters like Slow PAL.
The best combination I got is 1 repeated frame every 50th frame and 1 repeated frame every 25th frame>
But added a bunch of artifacts. and they work badly for the movie where is a lot of actions (it's a martial arts movie).

Let me explain my trick I found having done tons of research>
It's a kind of win but requires a lot of manual editing work (time) and power of my Mac mini.
I reached a half and my Mac became slowing down and failing to render. Though I have Mini in its max M1 configuration.

The pic with 3 read arrows shows my secret combination when I transfer the source video 29.97p to 24p timeline assuming that the initial source of the video was a film shot and projected in 24p (it was shot in 1,85:1) :On timeline, I use blue brick to mark combination 8:07 + 0:02 + 16:14 + 0:02 + 16:14 + 02 + etc.
It simplifies to mark in/out the every portion of the video on timeline to be cut.
I have to have 2 frames in between these "bricks" to apply switching a conform mode, from floor to nearest neighbour and optical flow
The most difficulty is in the 2 frames cut before each 1st 8:07 portion. I have to search and find a solution how to avoid strong artifacts caused by optical flow interference.
The 2nd pic shows where I'm now, red square shows the portion left, but my Mac can't be as powerful as I need for this job.
But any way,
I did a test prior beginning to cut the full movie and result was excellent, probably this manual mode can help as the last mode I should

apply - but it requires a lot of research and exchange to other frames on optical flows

Hi colleagues!
I don't know how but I've managed to put the source video (29,97p) to 23.98p, then I applied Nearest Neighbour conform mode and the frames became a constant flow. Just 1 cut I found inside (closer to the end and about 7 mins) which had copied frames time-to-time. I cut this cut and exchange with the same cut from another copy of the film, applied another mode - floor before, and the film became completely uninterrupted. The last milestone I applied was conversion from 23.98p to 25p, after Topaz Al-12. I lost 4% in the video duration (the film became a little bit faster, but it doesn't have repeated frames at all. Now I'm trying Amazon MediaConvert, will see how they convert to a European standard.
And please don't worry the job I'm making is legal. We bought licence for 5 Chinese films from an American company (it's why we got NTSC copies) but this company supplied us very bad let's say cracked versions of the movies made many years ago. Never buy movies from them.

I'm trying to transcode the 2nd movie, and got another combination of coupled frames: 3 + 3 + 3 + 15