CUDA Encoding vs. Handbrake

[pizzaboyUK]

I don't have the most powerful PC in the world, so when I heard I could use my GPU to speed up my encoding times, needless to say my eyes nearly popped out of my head!

Currently performing my first encode using MediaCoder, MKV H.264 with CUDA, constant bitrate with output size 2.1GB. Bit disappointed there doesn't appear to be an ETA (though the progress does seem a lot quicker) and also the summary shows the estimated size to be 1274MB - way off my specified output size...

Reading up a bit after the encode started, I discovered some posts stating CUDA comes up with quality not on par with normal encoding (though that was way back in 2010), and even on MediaCoder's website, they stated the output sizes were generally quite inaccurate.

Anyone used MediaCoder extensively like to share their experiences? And more importantly, should I just switch back to encoding with VidCoder/Handbrake?

[videobruger]

Use bitrate instead of targeting size. Encode 5 min test wih and without CUDA, compare them and judge the quality for yourself.

[jagabo]

If you have a quad core or better CPU, x264 at the veryfast preset will be faster than CUDA and produce better quality. On slower CPUs CUDA should be faster than the CPU. You'll have to decide for yourself whether the quality loss is worth the speed gain.

Intel's Quick Sync encoder on Ivy Bridge (3xxx series) or Haswell (4xxx series) CPUs is faster than x264 veryfast but delivers lower quality. Haswell's QS is faster and delivers better quality than Ivy Bride QS (or worst of all Sandy Bridge, 2xxx series). With Haswell the quality loss is low enough (compared to x264) that it's almost worth using for the extra speed.

[pizzaboyUK]

Left this running before getting some kip. Awoke to find a very annoying donate popup which could only be skipped via making payment or killing via Task Manager The encode appeared to be finished but the Temp file not converted to a readable format!

Despite the attraction of some of the programs features, no way I'm gonna put up with this BS!

[jagabo]

The last time I tried Mediacoder, a year or so ago, you could just close the popup and it would finish the encode. I think there was another workaround, blocking their server in your hosts file, or something like that.

[videobruger]

Workaround

https://forum.videohelp.com/threads/358944-Mediacoder-now-ransomware

[pizzaboyUK]

Thanks for the tip guys.

If a FREEWARE company puts so much effort to make their product annoying even before I'm able to test it, I'm not gonna bother to support them.

I'm sure if it was feature-rich, more popular programs will imitate them, if not - then probably those features weren't worth bothering in the first place.

[MaverickTse]

Thanks for the tip guys.

If a FREEWARE company puts so much effort to make their product annoying even before I'm able to test it, I'm not gonna bother to support them.

I'm sure if it was feature-rich, more popular programs will imitate them, if not - then probably those features weren't worth bothering in the first place.

This is somewhat unfair to developer. MediaCoder (and also a lesser-known "AudioCoder") has been around quite a while until the ads suddenly became annoying. Probably, so many people download it but users hardly ever donate/buy it (and perhaps of little value by writing this into his CV also)

Imagine you have made something that a million people "thanks" you but not getting a cent in this capitalist society, how would you feel?

Also note that it is not a "company" really, just another one-man project.

For the CUDA thing, there has been many reviews and tests and the conclusion is GPU-accelerated encoding cannot beat CPU-only encoding.
If you need speed and have an Intel Core-i, try handbrakes' QSV encoding. QSV, IMO, is faster than CUDA.
Think about on what device you video will be viewed. On handheld devices, probably you don't need BEST quality and fine with GPU-accelerated encodes.

[DarrellS]

I tried CUDA a couple of years back (or more) and came to the same conclusion as Jagabo. Not only was my Q6600 CPU faster at Superfast preset, the quality was much better and not only was the filesize of the CUDA file larger than the x265 file, it was bigger than the original file so it was a loser all the way around.

If you are headstrong about using CUDA. You can use the external encoder feature of Virtualdub with the Cuda.exe commandline encoder or Selur's Hybrid program which uses the Cuda.exe encoder.

Selur can help with his program and myself and a couple others here can help with Virtualdub. There is also a thread on Virtualdub forums with a tutorial for using the external encoder feature. Here are two files you could use. The first is a master file for most commandline encoders and the second is just for Cuda 264. Just change extension from .txt to .vdprof. Open Virtualdub > Options > External encoders > import > Cuda.vdprof . You need to either change the path to your encoders or move your encoders to the paths in the vdprof file.

[clamo]

the thing about GPU encoding is that it's meant for use with extremely HI bit-rates. 20k + bit rates for bluray and 4K encoding. using cuda on any lower than that will produce really bad looking video.

[pizzaboyUK]

the thing about GPU encoding is that it's meant for use with extremely HI bit-rates. 20k + bit rates for bluray and 4K encoding. using cuda on any lower than that will produce really bad looking video.

Thanks for that clamo. Probably the advice I was really looking for.

[pizzaboyUK]

This is somewhat unfair to developer.

Imagine you have made something that a million people "thanks" you but not getting a cent in this capitalist society, how would you feel?

Mav, firstly... MediaCoder should not be advertised as FREEWARE if its not.

Secondly, its a software frontend. The purpose of a frontend (or any other software for that matter) should make its users' use of time more efficient or more intuitive for whatever its purpose. Compared to other freeware software out there, it definitely is not to commercial standard for ease of use (that I would be willing to pay anyway)...


I agree with you to some extent that people generally think they deserve and demand more than they give, but I also don't condone the dev's method of trying to extort money. They have a choice - trial period... limited use... or just mass ads... if your product is good, people will buy it. I don't think you can really rely on donations.

My first use, without even managing to determine whether the software works or not and I'm confronted with a hostage situation. If your gonna promote anything in this world, I really would not recommend promoting bad experiences...

[MaverickTse]

This is somewhat unfair to developer.

Imagine you have made something that a million people "thanks" you but not getting a cent in this capitalist society, how would you feel?

Mav, firstly... MediaCoder should not be advertised as FREEWARE if its not.

Secondly, its a software frontend. The purpose of a frontend (or any other software for that matter) should make its users' use of time more efficient or more intuitive for whatever its purpose. Compared to other freeware software out there, it definitely is not to commercial standard for ease of use (that I would be willing to pay anyway)...


I agree with you to some extent that people generally think they deserve and demand more than they give, but I also don't condone the dev's method of trying to extort money. They have a choice - trial period... limited use... or just mass ads... if your product is good, people will buy it. I don't think you can really rely on donations.

My first use, without even managing to determine whether the software works or not and I'm confronted with a hostage situation. If your gonna promote anything in this world, I really would not recommend promoting bad experiences...

I agree that the ad is too invasive and I'd also think it is better for MediaCoder to go payware, but I can't agree with the rest of your arguments
.
Be it frontend or backend, programmer needs to read up API documentations, extensive test... may be it is easier than writing the backend, but still a lot of work to put everything to work.

Wikipedia also runs on donation, Blender also runs on donation... but only a handful of famous OSS will get decent financial support.

You can look at the Software list for transcoders on Videohelp. Most are either adware, payware or trial-ware

if your product is good, people will buy it

Nope. This is a BIG mistake.(I once had this misconception when I was young...)
The reality should be:
If there is a good product, people will try to rob it first.

Only when for some reason people cannot escape from a service/product, then people will consider buying it

I'd suggest you try learn some programming... and try writing something generally useful then publish it... then you shall understand the plight of a nobody OSS/freeware developers.

[pizzaboyUK]

A lot of software devs know how difficult it is to sell their products. The majority make software primarily for themselves first, then maybe share with the world if they think its going to help others. In that respect, any donations or "profit" is a bonus.

If there is a good product, people will try to rob it first.

On a more realistic level... if someone is going to steal your product - do you really think they were a potential customer who you were going to make a profit off anyway?

You could look at it on a more positive side... your product must have been good enough for someone to waste their time to hack it. You could look at it as promotion - if your product wasn't so "available", would you have got so much exposure?

Like any business, everyone has their problems to contend with - its something you can't avoid, and you must work your way around.


At the end of the day... the world is... as it is. As with any self employed business, if you can't make enough profit from what you make, the sad truth is... don't leave your day job.

But I digress, this is going in going into morality, ethics and business - out of scope for this dicussion.

[MaverickTse]

A lot of software devs know how difficult it is to sell their products. The majority make software primarily for themselves first, then maybe share with the world if they think its going to help others. In that respect, any donations or "profit" is a bonus.

If there is a good product, people will try to rob it first.

On a more realistic level... if someone is going to steal your product - do you really think they were a potential customer who you were going to make a profit off anyway?

You could look at it on a more positive side... your product must have been good enough for someone to waste their time to hack it. You could look at it as promotion - if your product wasn't so "available", would you have got so much exposure?

Like any business, everyone has their problems to contend with - its something you can't avoid, and you must work your way around.


At the end of the day... the world is... as it is. As with any self employed business, if you can't make enough profit from what you make, the sad truth is... don't leave your day job.

But I digress, this is going in going into morality, ethics and business - out of scope for this dicussion.

If I were a school kid, I would have agreed with your arguments. This world is more insane than you'd have think.
> Even for people who try to steal a product, if somehow manage to stop them from stealing the whole thing and there are no other alternative, you can force a portion of those thief to paid.(but for video encoding, there are MANY alternatives, just perhaps slower or less convenient)

> Hacked soft as promotion!? You should tell this MS and Adobe...

> it is not really a "business" thing, it is a matter of ethics and integrity... dev's(and charity in general) probably don't mind individual just donate $1, but no one will be happy (and not practical) if everyone just say "thanks", but rob the warehouse empty, and no body paid that $1.
That's why a lot of freeware got disbanded after a while. The devs either have to make a living, or simply fed up with user (irresponsible) response.

Well, you may not agree or understand what i'm saying. If someday you get into this kind of activity yourself, you will get it.