Will Re-Rendering A MP4 File Cause Quality Loss?

[Steve Rhoden]

They are not exactly the same. The re-rendered version
does suffer quality loss, but its just not visible to the
naked eye, because its so minimal.

Steve Rhoden
(Cow Leader)
Film Editor & Compositor.
Filmex Creative Media.
https://www.facebook.com/FilmexCreativeMedia
1-876-461-9019

[Jackie Luffy]

Thank you! I was also wondering as well

Say the original one time only rendered mp4 was 200 Megs

and I re-rendered it to purposly be bigger like 250 Megs

Since the bitrate was set higher would it keep the quality in tact and would it be worth it?

Or should I just try and match the bitrate so the re-rendered one will be 200 megs just like it was before?

[Steve Rhoden]

Since the bitrate was set higher would it keep the quality in tact and would it be worth it?
……No it would not and therefore would not be worth it.

Or should I just try and match the bitrate so the re-rendered one will be 200 megs just like it was before?
…..Exactly.

Steve Rhoden
(Cow Leader)
Film Editor & Compositor.
Filmex Creative Media.
https://www.facebook.com/FilmexCreativeMedia
1-876-461-9019

[Jackie Luffy]

Thanks for your help!

[Steve Rhoden]

No problem…

Steve Rhoden
(Cow Leader)
Film Editor & Compositor.
Filmex Creative Media.
https://www.facebook.com/FilmexCreativeMedia
1-876-461-9019

[John Rofrano]

[Jackie Luffy] “For years, Ive always known that re-rendering ALWAYS will lose quality!”

I just want to correct your original statement because it is not always true.

There are codecs that are lossless (e.g., Lagarith, HuffYUV) and you can re-render 1000 times and there will be NO LOSS of quality. So saying that you will ALWAYS loose quality is not true. Only “lossy” codecs will loose quality; “lossless” codecs will not. Also rendering “uncompressed” will not loose quality either.

MPEG4 AVC/H.264 is lossy so your statement is true that re-rendering to MP4 will always loose quality even if it is “visually” lossless as you have seen.

People are always worrying about quality when re-rendering and most of it is unfounded. I’ve worked on projects that were shot in HDV and we’ve re-rendered sections back to HDV and dropped them back onto the timeline and delivered the show for national broadcast a no one could see any difference from the original HDV footage and the footage that was rendered twice so don’t angst over it.

My general rule of thumb is: “if you can’t see the difference, there is no difference” 😉

~jr

http://www.johnrofrano.com
http://www.vasst.com

[Dave Haynie]

[John Rofrano] “People are always worrying about quality when re-rendering and most of it is unfounded.”

That’s basically it.

Every time you render MPEG-2, MPEG-4, or any other of the family of Direct Cosine Transform CODECs, there’s a conversion process. The pixel data is bucketed into macroblocks (usually 8×8 or 16×16 pixels, but particularly in AVC, that’s not the whole story). Then the DCT is run,w which translates spatial information into frequency information. This step isn’t the lossy part.

Next comes the lossy part — a low-pass filter is run over the frequency data, tossing out some of it. Sometimes. The bitrate, some of the VBR analysis of the “big picture” if you’re encoding variable bitrate, etc. all contributes to shaping that low-pass filter. After the low-pass filter, there’s additional stuff: lossless Huffmann or similar compression of the frequency data, etc.

When you recompress with the same or greater compression settings on identical material, you’re pretty likely (but not guaranteed) that the AVC encoder is making the same blocking, bit-budget, and filtering decisions. Which means the video is going to change very little. And of course, you probably did something to the video in the process of editing: color correction, noise reduction, level or contrast adjustments, etc. That’s going to change the encode-time decisions, but if you’re doing your edits well, the bottom line is that the re-encoded video looks better than the original video.

And these days, we can often start with much higher bitrate video than we target, whether that’s DSLR video at 50-100Mb/s, AVC-Intra or some other higher-end formats, etc. At this point, you’re going to be essentially indistinguishable from RAW at the target video, aside from any processing errors (letting video saturate and clip, etc… most of which can be avoided by rendering with 32-bit precision). Given the higher quality video, you can also do things to ensure that the final encoding preserves more of the intended video, less of the bad stuff — like light noise filtering, if you have a little digital noise in the image.

Back in the early days of compressed music, one of the figures of merit on the CODEC was the number of re-compressions you could stack without some level of damage to the audio. I recall, when MD first came out, Sony’s ATRAC encoder could manage about 4-5 recompressions; by the time they were done editing with it, it could do 20-25.

-Dave

[Ronald Puerto]

Pro editing programs do have methods where the video is not re-encoded. Sometimes it can be hard to tell, the rapid speed which it writes the new file out though is one clue. You can also cut usually also and have no re-encoding occuring. Some key things are to make sure your project settings, the input files and the output settings all match. Some of them even have a “settings viewer” for that purpose. ANY filtering, adjusting, transitions added will definitely be re-coded. It can also be very important what codec is being used, the program just might not support that for some codec. For more information read this post.