Rendering Quiz

David Laine wrote on 5/29/2014, 10:55 AM
Hi all

SWMBO is on my case to get on with putting hours and hours of cat and model railway footage we have shot on our AVCHD camcorder into something like order.

I have read on here and other forums that GPU encoding looks less good than CPU encoding.

After a lot of testing and borrowing video cards the fastest renders seem to come from Intel Quicksync on a i73770 HD 4000.

I have looked and looked at the QS and CPU renders and I cannot see any difference on our 42" plasma TV

So I have an idea, I have uploaded 2 files one is rendered on CPU the other Quick sync, I have added a few transitions and titles to the footage.

Just to be fair to both render types I have called them A and B so hopefully viewers will not have any preconceived ideas as to the better render type

Can I ask you folk here to download both files and see if you can see what is Quick sync and what is CPU

If you do see the differences please let me know what to look for

I will post the answer to what is A and B here in about 2 weeks

Here are the links to both files

http://flashbacksales.co.uk/test/quiz%20a.m2ts

http://flashbacksales.co.uk/test/quizb.m2ts

BTW the edit programme is V13 64bit PC is running W7 64bit

May thanks for your help with this

Comments

videoITguy wrote on 5/29/2014, 11:02 AM
Lets say I detect technical differences in your samples which can easily be proven. SO what?

The difference that will matter to you is determined by several issues. One of which you touched on is 'expediency". It is a factor in production workflows.
So yes you have technical differences, but what will it matter to your eyes and those of your audience who view this. Don't fret about this.
Tech Diver wrote on 5/29/2014, 11:26 AM
In reading the OP, I was immediately stumped by what SWMBO could mean. I had never seen this before and tried to guess some highly technical acronym like "System Wide Mega Binary Overload" or perhaps "Synchronous Web-like Micro Biphasic Oscillation". After a Google search, I was rather disappointed to learn the true meaning.

Anyway, if you can't tell the difference between the two rendering approaches without extensive analysis, then it doesn't matter whatsoever.

Peter
PeterDuke wrote on 5/29/2014, 5:49 PM
SWMBO will be familiar to old Rumpole viewers
john_dennis wrote on 5/29/2014, 8:52 PM
I watched the videos at work using Windows Media Player from Windows 7 on a "corporate" PC and I could not tell the difference. A casual visitor to my office couldn't pick a favorite, either.

On a 43" plasma TV the differences are even less apparent.

I ran an analysis of the two streams and there are very small visible differences using the analysis tools, but I made no attempt to determine which of the videos is which.

Visually, I'm neutral. If I was forced to pick one over the other, I would throw a dart or I would take B since the file is a little bigger or A since the file is a little smaller.

About a year ago, I stopped using GPU to render because for me It's not worth the hassle.

My past tests with time and files size results are in this thread.

The most fitting abbreviation from my point of view is FFIAWT.
musicvid10 wrote on 5/29/2014, 9:37 PM
You are asking us to compare two different renders to each other, rather than comparing each to the source.

Only possible way to quantify something like this is with the original source as the reference, that way comparing things like SSIM and PSNR would become relevant.
David Laine wrote on 6/6/2014, 9:18 AM
Hi

Thanks to those who replied to my original post

I was only hoping to be told what I might not have noticed in the HD4000 render

BTW the GPU render was B

Sorry about using the term SWMBO I forgot this is an international forum, here in the UK it is widely known. Bit like 'er indoors' that was made famous in the uk tv show Minder. I believe that saying has made it's way into the Oxford dictionary.

Anyway the software is all clear to me now - time to do ask asked by 'the wife'