Windows Video encoder for compressing large videos

S

sauravghosh

Level G
Finally, I've run out of the 15GB storage Google offers. It's at 98% now and I must compress some of the videos and transfer it away from Google photos to somewhere else. These are mainly all documents of which I take video records as evidence just before submission (i.e. docs submitted in medical insurance, in court etc.).

Presently I use Handbrake and simply use the default 22 value for conversion. Anything better? Retention of highest quality at the smallest size is the goal here obviously.
 
Are these orginal quality files or have they already been transcoded (by you or by Google)? Also, what encoder settings do you use: codec, encoder speed, frame rate, etc? x265 instead of x264 and dropping the frame rate would get you some easy savings right out the gate. You could also use AV1, which gets you really good results, but takes a really long time.

Also, are you sure you want to spend time doing this? You'd probably be better off buying an SD card to store the backups, with one or two extra for redundancy if you need it. This would have the added bonus of not risking loss of detail through transcoding. Unless you need the accessibility of the cloud, that's what I'd do.
 
For documents, photos provide way better quality at a fraction of the file size. Video increases file size and gives you lower quality.

If you're doing this for legal reasons, your re-encoded video might not hold the same weight as the original camera file, as this could be considered altered.

Your easiest option may be to create another account or buy more cloud storage. With single external storage on SD/HDD you will lose data if device fails. A couple of 1TB HDD will let you store a lot of stuff
 
SirVer said:
Are these orginal quality files or have they already been transcoded (by you or by Google)? Also, what encoder settings do you use: codec, encoder speed, frame rate, etc? x265 instead of x264 and dropping the frame rate would get you some easy savings right out the gate. You could also use AV1, which gets you really good results, but takes a really long time.
Click to expand...
These are original. Taken with my mobiles directly and uploaded to G photos.
As for Handbrake, I never bothered to check settings. I simply drag and drop the video and hit convert RF remaining at default 22, that's all.
n1r0 said:
For documents, photos provide way better quality at a fraction of the file size. Video increases file size and gives you lower quality.

If you're doing this for legal reasons, your re-encoded video might not hold the same weight as the original camera file, as this could be considered altered.
Click to expand...
Altered metadata is fine. Still photos are always taken. Videos are additional measures as they are better proof than a still photo.

I would like these to be on cloud for accessibility, otherwise hard disk would have been fine. SD card I don't trust, they are slow and get corrupted often. Besides i don't remember any phones i had in past few years supporting it.
 
How quickly are you filling up this space? If it took years to get to 15GB, then I'd probably transcode, as you can often compress your average camera output to 30-50% of the orginal size. However, if it took less than a year, I'd probably pay for the extra storage (₹1300/year for 100GB from Google), since you'd have to do that soon enough anyway. You can also always transcode later if you want to.

Btw, if you do want to try transcoding, check if your videos were uploaded to Google Photos in "original" quality; IIRC, if you don't explicitly enable that option,Google automatically transcodes the video to save space, and transcoding a video that's already been transcoded can easily result in massive quality loss.
 
Thanks all. I asked whether there's any better software than Handbrake for the job. How to compress my original large videos to small size in the most efficient way, if there's any input on the subject I would appreciate that. Otherwise I think Handbrake method is fine with me.

I won't pay for any cloud storage with their recurring subscription model especially offering tiny space. 1300/yr for 100 GB is ridiculous, will never consider it.

@SirVer These are all in original quality, I made sure of that. Also it took me about 10 years to fill up this space, although my Gmail ac is about 16 years old but I didn't upload anything in the early days.
 
sauravghosh said:
I asked whether there's any better software than Handbrake for the job. How to compress my original large videos to small size in the most efficient way
Click to expand...
Most transcoder GUIs use ffmpeg as the backend (including Handbrake), so the actual software doesn't matter, the settings do. As for what settings are best, it's highly dependent on several factors, including:
  1. Source resolution and bitrate
  2. How static the video is
  3. Noise/grain in the video
    1. Common occurrence when lighting is less than ideal, especially with older cameras
  4. Fineness of detail that needs to be preserved
    1. This would be influenced by how the videos were shot: recorded from a regular height and flipped page by page like in document scanning, or up close moving down each page, or something else
It's therefore difficult to say what would be ideal as I don't know what the source looks like, especially since this is far from a common use case. As I mentioned before, you can start by switching to the x265 encoder, reducing the framerate to a divisor of the original (eg. 60fps to 30fps to 15fps), and changing the encoder preset to the slowest you can tolerate. You can leave the CRF at 22 (if source is 1080p, 18 if lower) to start with and check if it's preserving the detail you need, and go from there. Be aware that if your source files are 1080p and in the range of 30-40MB/minute (i.e. 4-6Mbps) you may not be able to reduce the file size much without losing detail. Also be sure to test your settings across multiple different samples to ensure they scale well to your entire dataset.

Finetuning all of these things can take hours or even days (depending on encode time and dataset size), and that's before getting into more resource intensive codecs like AV1 or preprocessing filters and so on - this is why I suggested alternate storage options. And to be frank, using video to do backups seems like a less efficient and less effective method compared to the photographs you've taken, and I can't imagine a context where those would be less valid than video taken by (presumably) the same camera at the same time, especially if altered metadata is acceptable; that said, it's your workflow, your choice. If you do transcode it all, I strongly recommend keeping a backup of the originals on local storage - burn them to a few DVDs, copy them to a couple of flash drives, anything.

P.S. You can also use try the 10 bit version of the x265 encoder, since you can sometimes get better results even for 8 bit content, but you may run into compatibility issues with Google Photos and/or certain devices.
 
sauravghosh said:
Thanks all. I asked whether there's any better software than Handbrake for the job. How to compress my original large videos to small size in the most efficient way, if there's any input on the subject I would appreciate that. Otherwise I think Handbrake method is fine with me.

I won't pay for any cloud storage with their recurring subscription model especially offering tiny space. 1300/yr for 100 GB is ridiculous, will never consider it.

@SirVer These are all in original quality, I made sure of that. Also it took me about 10 years to fill up this space, although my Gmail ac is about 16 years old but I didn't upload anything in the early days.
Click to expand...
Setting up a long-term Microsoft 365 sharing group over here was one of the best things I did. Spending 650/year for 1.1 TB OneDrive space along with all Office products is really a bargain.

That said, if you have a recent Nvidia GPU and don't want to spend ages finding something that provides that 5% better quality, NVENC HEVC encoding on Pascal or newer architecture is good enough.

NVENC Application Note

docs.nvidia.com docs.nvidia.com
 
@SirVer Yes thanks. You already mentioned some parameters, any I'll test them out when I get the time. My msg was not meant for you as you already replied on topic.

I just noticed one thing about Google photos. You see I take a photo from distance and then crop out the irrelevant. The cropped photo is of much lower size, obviously, but G Photos is still backing it up in full size (before the crop). Presumably because a user might want to retrieve it later. Is there any way to make the change permenent? I don't need the full photo. Attached a screenshot for clarification in case my query is unclear. Notice the difference in size bw on device and backup.
 

Attachments

  • Screenshot_20230830-175026.png
    Screenshot_20230830-175026.png
    296.5 KB · Views: 64
  • Screenshot_20230830-172638~3.png
    Screenshot_20230830-172638~3.png
    135.5 KB · Views: 62
t3chg33k said:
NVENC HEVC encoding on Pascal or newer architecture is good enough.
Click to expand...
What settings are you using? I've experimented a lot with NVENC HEVC on both Pascal and Turing, and most of the time the results end up larger than source when using constant quality encoding (the only exception being anime). It does work well when I use average bitrate instead (which makes sense as it's meant for streaming), but I can't get it to give good results with constant quality no matter what I try. I also observe significantly less detail reduction and blocking with x265, which is why I recommended it instead of NVENC. If there's something I'm missing I'd love to know - having to use CPU for everything is a massive pain.
sauravghosh said:
Is there any way to make the change permenent?
Click to expand...
You can save as a copy instead of saving in-place when doing a crop, but I'm not aware of a way to discard the original once it's already saved in-place. The only way I can think of is to download them all, delete the originals, then reupload them.
 
SirVer said:
What settings are you using? I've experimented a lot with NVENC HEVC on both Pascal and Turing, and most of the time the results end up larger than source when using constant quality encoding (the only exception being anime). It does work well when I use average bitrate instead (which makes sense as it's meant for streaming), but I can't get it to give good results with constant quality no matter what I try. I also observe significantly less detail reduction and blocking with x265, which is why I recommended it instead of NVENC. If there's something I'm missing I'd love to know - having to use CPU for everything is a massive pain.

You can save as a copy instead of saving in-place when doing a crop, but I'm not aware of a way to discard the original once it's already saved in-place. The only way I can think of is to download them all, delete the originals, then reupload them.
Click to expand...
You are right, quality is not the forte for fixed function encoders. But since the OP simply wants to store documentary videos, quality is not critical whereas time and space might be, hence the recommendation.
 
SirVer said:
You can save as a copy instead of saving in-place when doing a crop, but I'm not aware of a way to discard the original once it's already saved in-place. The only way I can think of is to download them all, delete the originals, then reupload them.
Click to expand...
Ah that's it, thanks a lot. Wish that option was available for existing photos though. But I'll make sure to use that option from now on.
 
t3chg33k said:
But since the OP simply wants to store documentary videos, quality is not critical whereas time and space might be, hence the recommendation.
Click to expand...
I was thinking the opposite - since there might be a lot of fine print on these documents, retaining quality would be more important than usual. Would depend on how OP has recorded the videos I think.
 
I tried the AV1 encoder last night. Took 3 times longer than h264 as you said but problem is it wouldn't play in any player built into windows 10 - Movies, WMP and even the MPCHC which is what I use as my go to. So I'm guessing this is a new codec and compatibility might be an issue.

I did install the required codec anyway, but in the end was exhausted to compare the quality with h264 but I'm sure it was better.

@SirVer
 
sauravghosh said:
I tried the AV1 encoder last night. Took 3 times longer than h264 as you said but problem is it wouldn't play in any player built into windows 10 - Movies, WMP and even the MPCHC which is what I use as my go to. So I'm guessing this is a new codec and compatibility might be an issue.

I did install the required codec anyway, but in the end was exhausted to compare the quality with h264 but I'm sure it was better.

@SirVer
Click to expand...
It is a relatively new codec, but most recent devices have had hardware level decoding for it for years now, and both MPC-BE and VLC open it just fine for me without any additional codecs. If you want a faster encode, try H.265 - it's not as efficient, but in many cases it's good enough.
 
Back
Top