Changing pre-rendered video footage using csv data

[Tomas Bumbulevičius]

Hey Perry, to begin this – this could be solved relatively easy with scripting, but I have one question:

Considering segment 1 has 4000 options, and segment 3 and 4 contains even more customisations… Wouldn’t it turn into 4000x120x15 variations (blows my mind thinking about results, haha), if videos suppose to be unique? Elaborating further on what you have in mind for those segment-variations would be a good idea to help you out better!

Find out more:
Motion Graphics Design & After Effects Tutorials
On YT
On VH

[Perry Sheppard]

Hi Tomas,

Thanks for your reply. Permutations and combinations abound lol, but, no, it’s just 4,000.

There are 4,000 rows of data and each row contains the information for one individual. Their name, location, and area of interest. There are 4,000 ‘people’, 120 ‘locations’ possible for each individual, and 15 possible ‘areas of interest’ for each individual.

It’s essentially a video mail merge:
Segment 1: The person’s name. This would appear as dynamic, animated text (provided by the csv) in the comp.
Segment 2: Common to all
Segment 3: The person’s location. This would appear as a pre-rendered video specific to the location (video footage to appear would be driven by the csv data).
Segment 4: The person’s area of interest. This would appear as a pre-rendered video specific to the person’s area of interest (video footage to appear would be driven by the csv data).
Segment 5: Common to all

To map out the comp, it would be something like:
Segment 1: Text animation – 8 seconds (text dynamically changes via csv)
Segment 2: Pre-render into – 10 seconds (static)
Segment 3: Pre-render location – 8 seconds (footage dynamically changes via csv)
Segment 4: Pre-render area of interest – 15 seconds (footage dynamically changes via csv)
Segment 5: Pre-render outro – 4 seconds (static)

Thanks again!

[Tomas Bumbulevičius]

Perry, thanks for reassuring me down to 4,000 videos only, haha! This all makes a lot more sense and clearer in general.

Now what is important here, is how you are going to do that ‘linkage’ between pre-rendered assets and the .csv.
1. Will it be based in some way on file names, or otherwise?
2. Assuming files replacement is sorted – what is your rendering preference?
3. Is it a single-outputs generation only, or might be further updated in the future with other data sets?

Cheers!

Find out more:
Motion Graphics Design & After Effects Tutorials
On YT
On VH

[Perry Sheppard]

Thanks Tomas,

1. Re: linkage to csv – I need to refine the data set regardless of what approach I take (it’ll only take a few minutes to do), so I have options. I was thinking my best options would be adding a column in the csv for, example, ‘Location’ with content in the format of either ‘United Kingdom.mov’ or simply ‘United Kingdom’. Whichever is most efficient. The comps would ideally be named to match with the originating csv document. Ex. row 2 = comp 2.

2. Re: rendering preference – are you referring to render engine, settings, both? If engine, then either AE Render Engine or Media Encoder. I know that aerender would probably be faster, but I’m not familiar with Terminal and wouldn’t be comfortable using a solution I couldn’t troubleshoot. For output settings, it would be mp4, 1080×1080… I’ve done a fair bit of research on maximizing render efficiency (source file colour space, codecs, etc.), but always open to new info.

3. Re: one-off or not – Ideally, the setup would work with a new data set (the new data set would have exactly the same structure though) and new comp materials. So, the content of the csv rows could change, but the columns would remain identical and the master comp might change, but the approach would remain consistent.

Let me know if I understood your questions. Thanks again 🙂

[Perry Sheppard]

One additional nice-to-have would be to also be able to render out the first second of each comp as an animated gif. I know that media encoder can export these, but would either need an in point at 0 time and an out point at 1 second, or the length of the comp would need to be one second.

[Filip Vandueren]

Have you taken a look at templater from Dataclay?

[Perry Sheppard]

Hi Filip, hope you’re doing well. Thanks for this.

I just took a quick look at Templater and I’m already wishing I’d known about this two months ago. They need to bump up their SEO/marketing because it looks great, but this is the first I’ve heard of it in all of my searches (including searches on batch, render farms, etc.). I’m going to do a deep dive into the demo this weekend and see how I might be able to incorporate this into the workflow.

[Perry Sheppard]

Templator looks great, but I’m going to need to get this first project out as a proof-of-concept before I can try to make the case for it. At 3-6k/year, I’ll need buy-in on consistent use.

[Filip Vandueren]

I thought you could get a 1 month subscription?
Have not used it myself, but if I had a client ask for this kind of project with 4000 movies, I wouldn’t hesitate

[Perry Sheppard]

The trial is fully open for 30 days–which is great–but videos include a watermark. My proof-of-concept is going to require the production of the videos, but also the delivery to the recipients via email; otherwise, it won’t be considered.