Bitmovin has given us one million output minutes. For VOD, we currently use per-title encoding. We have seen that we have utilised 335k output minutes in the last two months, which is a quite substantial consumption. Would you kindly advise us on ways to reduce costs, or could we switch from per-title encoding to bitladder encoding?Because bitmovin materials say that per title encoding is cost-effective and improves video quality, we have opted for per title encoding over bitladder.
One more question is whether we can limit the number of resolution copies per title encoding.
Hi @vinayak.baye
Thanks for reaching out to us.
Can you please DM me your OrgID or your email address? I will take a look more precisely.
Per-title encoding generally aim to generate a more optimal amount of renditions for each content, which in most cases can result in savings (compared to a standard ladder), but can sometimes lead increased cost for a more complex content, in case a larger amount of high quality renditions is generated.
For each encoding job, you can open the Billing tab (right of the screen) to get access to the detailed billing minutes calculation and understand more your costs.
To save costs, you could try a few adjustments : Cap the max bitrate in the Per-title configuration
Increase the step size ( minBitrateStepSize / maxBitrateStepSize) : this should result in a bigger distance between each rendition, and therefore, less renditions and less cost overall. The downside is that switching from one quality to another might be more visible on the player side, and you risk to deliver a less optimal bitrate to certain users (eg. if one rendition is 500kbps and the next one 2Mbps, then, a user with only 1.5Mbps of bandwidth will only be able to play 500kbps if there is no renditions in between).
Thanks for your feedback.
I took a look into a couple of encoding jobs :
From what I see, Per-title results in generating just a handful of renditions, and your bitrate ladder seems nicely tailored to your content for the few cases I looked at, with a nice distribution across SD and HD renditions.
For HD, the coefs are higher for the HD (>=720p) renditions so the cost is slightly higher.
In summary, your setup is already pretty optimized and cost-effective. Optimizations around the step size is the only option I see and could possibly help you skip a rendition or two, and save 20%-25% of the cost, but it comes with some downsides, as explained before, so I don’t really recommend those.
At a high level, per-title already seems to work well for your use-case from a cost perspective : reducing the bitrate on a lot of your content, reducing the amount of renditions, adapting the image size to your content, and therefore not incurring the 2x multiplier on the HD renditions in a lot of case. I don’t think there is much room for improvement, on the cost side, and the cost benefits of per-title very likely outweighs its cost in your case, so my recommendation would be to keep your workflow as is.
