Solved Intermediate rendering format / codec choice?

Hi to all,
I am using a series of project video clips that are in AVCHD (m2ts) source format.
Once the project is complete the clips will be brought together and finally rendered in either Sony AVC (either mp4 or m2ts) or Main Concept (MPEG2).
Before that happens, the original individual short clips have to be rendered with an intermediate format allowing me to bring together what has been done in the project - too many overlays to leave as one time line development.
ie A video clip has ten overlays associated with it - that clip is then rendered with the resulting rendered video being opened up to allow another ten overlays to be added. Possible that this has to be repeated 3 or 4 times.
The potential for visual degradation is obvious!
So, any suggestions of which intermediate format / codec to use to limit this occurring would be gratefully received.
Many regards,
Peter

Avid DNxHD is one option.
John Rofrano has a tutorial on this topic which will take you through all the steps here.
johnrofrano.com/training/video-tutorials/avid-dnxhd-template-for-vegas-pro/

Went for a cup of Yorkshire's best - come back and there is the answer!

That, plus it rained today rather than snowed - makes one glad to be alive!

Will let you know how things go.

Many thanks Derek

Peter

Hi Peter,

I have used the DNxHD codec, in Movie Studio 13 Platinum and it is a good intermediate render option. One of the concerns I had at the time was the Avid DNxHD codec wraps the video in Apple .MOV container. Movie Studio and Vegas Pro do not like the .MOV container on a PC, because Quicktime only 32 bits, even though it is being used on a 64 bit machine, it can only access 2 gb of memory.

It runs in SMS13P and Vegas Pro fine, the problem is when too many clips are added to the timeline, then weird things happen like crashing, freezing, etc. From what I read problems could occur with as few as 20 events (I personally doubt that figure) to as many as 100-120 events.

I used it for short video for several months without a problem, before I decided to test it myself before starting a larger project. I got a freeze in around 80 events being added to the time line. Then I read a post stating that the limit seems to be the number of MOV events on the timeline, but not the size. So, as a test, I put 25 events on the timeline, rendered them out as intermediates (DNxHD 220X in a MOV container). I did this 5 times. I then added the 5 intermediate files, each containing 25 MOV clips), to the timeline with success. So in effect I had 150 MOV events on the timeline, 70 more than I could load individually, with no freeze or crash. I then made some FX, transitions, and color correction and rendered it all out to a final DNxHD file, with no problem.

If you like the results of intermediate DNxHD files, and come across some freezes or crashing, I wanted you to know there is a work around. It appears the number that will freeze any particular PC is really based on those particular PC's and the workflow of the owner.

I have since went to Vegas Pro and use .MXF files as intermediates. Of course Movie Studio does not support MXF. However, my main camera records in either ProRes 422 HQ, or DNxHD 220x, both in a MOV container. So, rather than fight the limiting issue of the Quicktime 32 bit Windows version, I just render intermediate files before I start a project.

Hope this helps and doesn't confuse or raise any alarms.

Hi George, Derek,
Before down loading the codec, I did a quick search on the web and soon came up with comments regarding the topic that you alluded to.

I also was wondering about it being usable in MS ver.11, it being 32bit (and here is me using it on a 64 bit machine!).

Still - in for a penny as the wise ones say.
Downloaded, installed and set up a custom template.
That indicated that 59.94p @ 1080 is not supported but 720 is - (alternative being 1080 @29.97).
Also realised from the choices that huge resulting files were going to be generated if I was unwise in setting bitrate.

Anyway - wanting to see, quickly rendered at 59.94p/720 @ 100mbs for a 2 min 35 sec video with 9 clips on the time line.
The rendering took - just under 5 mins and the resulting file was 1.85Gb (original material being 405Mb).
I realise that 100Mb bit rate is somewhat over the top since the original video has a bit rate of 26Mb! - but the 100 was the lowest choice for this occasion.

Regarding the potential problems re: crashing etc - excellent that you have succinctly put it into a format that I am able to copy / paste and add to my ever growing notes (what did you expect due to you being so wise!).

In this case - it will not be a problem for me since the # of clips will be fairly small.

To Derek - As you will gather from reading the above, all appears to be working out well. Keyframing is slow as I have to place one every 10-20 frames since all that can possibly move is doing so (Dog, camera and operator)- relative movements are not linear nor predictable - did you say previously there was another way of getting over this? I certainly do not consider the work as a chore, in fact quite relaxing to be honest.

Enough said - once something respectable is finished will put up on YT so that others may see what is possible even though the topic may not interest.

Peter

Hi Peter,

It appears you would prefer to render to 1080 at 59.940 fps progressive. If that is true, you might want to try this.....

Set the frame rate in the template at '59.940000' and the field order as 'None (progressive scan)'. In the DNxHD configure window, resolutions select '1080p/29.970 DNxHD 45 8-bit'. That should held a rendered intermediate at 1920x1080, progressive, 59.940 fps constant YUV 4:2:2 8-bits at a bit rate of 90 Mbps.

Even though this configuration is for 45 Mbps it is yielding 90, because we changed the fps from 29.970 to double that at 59.940.

I don't have any original 1080p 59.940 material to test this, but Mediainfo is showing this should work. If it does work, a 90 Mbps 4:2:2 8 bit file should work well your intermediate as it is much higher quality than your original. Of course this doesn't add quality to your original, but should yield a file with very little degrading. That is assuming your original is 1080p, 59.940 4:2:0 8 bit.

Yes the files are going to be larger than your originals, as using these intermediate codec weren't really envisioned to make rendering from lower bit rate files, rather larger ones, but still it works and has a distinct advantage I think for your purpose.

Also, another alternative for intermediate files is CineForm in an AVI container. However the CineForm render is going to be larger than the DNxHD I believe.

Hi George,
If you say that altering the frame rate in the template section holds and actually renders as such - very good.

I presumed that would not be accepted and it would default back to the setting indicated. Shall see.

The bit rate (driving the file size) isn't too much of a constraint in this case but would be taken into account in future.

1080p over 720p choice is been driven mainly by wishing to remain constant with source resolution, 59.94 frames because I will be slowing down the final out put to around 10% of the original speed. Even though starting with 59.94 fps isn't huge - the clarity and flow is fine for what I need to view.

Enjoying the journey!

Many regards,

Peter

I'm looking forward to see the results you get. Good Luck with your project. Figuring out the best workflow is a journey and hopefully yours will be rewarding.

Can't make mind up due to inexperience.
May be in makes no difference. - but my gut feeling is that there is!

So Project 1 is completed.
Now to continue it's :
Possible to render project one to Avid DNxHD. Then use that material to place new key framed overlays on the resulting video clip to produce project two
Then repeat the above another two times -
i.e Original material source format and used to create project one

---

render using Avid DNxHD

---

use rendered Avid DNxHD as base for Project two - then render that to Avid DNxHD again

---

then use that as base for Project three

---

render that with Avid DNxHD

---

again use that as base for project four

---

render that into final format.

or:
Original material used source for project one as above

---

Save as project one.

---

Open up Original source into a new project and keyframe new material

---

Save as project two

---

Open up Original source into a new project and key frame new material

---

Save as project three
Open up Original source into a new project and key frame new material

---

Save as project four.
Therefore four individual projects created.
Render project one to Avid DNxHD.

---

use result to render project two to it.

---

use that to render project three to it and finally use result to render project four to it.

The difference I see between the two work processes is :
1. The first means adding sequentially new material to previously rendered video clips.
2. The second method means adding material to un-rendered video clips (Original format material) and then sequentially rendering.
Plus:
Would the second method alter the cast of colour due to progressive overlaying?
I may well be talking a load of mushrooms.
Any comments before I dive in the deep end or are you all going to tell me (politely) to just go and trial each method and let you know the result!
Regards,

Peter

Hi Peter,

As far as rendering and re-rendering, it appears method one and two are the same. There would be a difference on the timeline between method one and two. If I understand your two workflows, correct me if I'm wrong.....

In method one, the 1st project you lay down the right front paw placement with a colored circle and keyframe that, to remain at the same place on the ground, as the camera and dog move along the scene until it is out of the frame.

In project 2, you lay down the left front paw placement with a different colored circle and keyframe that, to remain at the same place on the ground, as the camera and dog move along the scene until it is out of the frame. You are using an intermediate render of project 1, so the colored circle from the right paw placement is also in the event preview.

In project 3, you lay down the right rear paw placement with a different colored circle and keyframe that, to remain at the same place on the ground, as the camera and dog move along the scene until it is out of the frame. You are using an intermediate render results of project 1 and 2, so the colored circles from the right and left front paw placement is also in the event preview.

Finally, In project 4, you lay down the left rear paw placement with a different colored circle and keyframe that, to remain at the same place on the ground, as the camera and dog move along the scene until it is out of the frame. You are using an intermediate render results of project 1, 2 and 3, so the colored circles from the right and left front paw placement, as well as, the right rear paw placement are also in the event preview.

By the time you get to project 4, you already have three other colored circles keyframed along the saved path, and I would imagine the frame (preview) is get a bit cluttered.

With your method two, each project will start clean in the frame (preview), and I would think it a bit easier to concentrate on the paw for that project and keyframe the appropriate colored circle. Method two requires an extra step for the 2nd, 3rd, and 4th projects as you need to replace the original event with the intermediate render of the previous project before that project gets rendered. This may be six of one half dozen of the other. If you have no problem keyframing the circles, as you go along when they start to build up from previous projects, I would go with method one, otherwise method two provides you an uncluttered preview.

Peter wrote: ~~~ or are you all going to tell me (politely) to just go and trial each method and let you know the result!

.....yep! That's what I would do.