File Size Quadruples After Rendering

Sweet Charles wrote on 12/18/2014, 12:59 PM
I wish to keep my same FPS which is 59.94. I also want to keep it at 1920 x 1080 P.
It appears to me I have only 1 choice by Sony to closely match my video which is the
XAVC S HD 1080 Long GOP 59.94 P without giving up anything.

Note: I am already putting in a compressed video.

[Video Information]
Codec:AVC1 - Built-in FFmpeg Decoder(h264)
Input type:AVC1(24 bits)
Input size:1920 x 1080(1.78:1)
Output type:YUY2(16 bits)
Output size:1920 x 1080(1.78:1)
Frame rate:59.94
BitRate: Unknown

[Audio Information]
Codec:AAC(0xaac0) - FFmpegMininum64.dll(aac)
Sample rate:48000 -> 48000 samples/sec
Bits per sample:16 -> 16 bits/sample
Channels:2 -> 2 channels
Bitrate: 254 kbps
General
Complete name : K:\2014-08-07\Ray of Hope\036.compressed.mp4
Format : MPEG-4
Format profile : Base Media
Codec ID : isom
File size : 188 MiB
Duration : 1mn 53s
Overall bit rate : 13.9 Mbps
Writing application : Lavf55.34.100

Video
ID : 1
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L4.2
Format settings, CABAC : Yes
Format settings, ReFrames : 4 frames
Codec ID : avc1
Codec ID/Info : Advanced Video Coding
Duration : 1mn 53s
Bit rate : 13.5 Mbps
Width : 1 920 pixels
Height : 1 080 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 59.940 fps
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Bits/(Pixel*Frame) : 0.109
Stream size : 184 MiB (98%)

SONY XAVC S HD 1080 Longg GOP 59.94 P

[Video Information]
Codec:AVC1 - Built-in FFmpeg Decoder(h264)
Input type:AVC1(24 bits)
Input size:1920 x 1080(1.78:1)
Output type:YUY2(16 bits)
Output size:1920 x 1080(1.78:1)
Frame rate:59.94
BitRate: Unknown

[Audio Information]
Codec:twos(0x736f7774) - PCM
Sample rate:48000 -> 48000 samples/sec
Bits per sample:16 -> 16 bits/sample
Channels:2 -> 2 channels
Bitrate: 1536 kbps
General
Complete name : K:\2014-08-07\Ray of Hope\sony xavc s hd 1080 log gop 59.94p.MP4
Format : XAVC
Codec ID : XAVC
File size : 173 MiB
Duration : 21s 522ms
Overall bit rate : 67.3 Mbps
Encoded date : UTC 2014-12-12 20:28:32
Tagged date : UTC 2014-12-12 20:28:32

Video
ID : 2
Format : AVC
Format/Info : Advanced Video Codec
Format profile : High@L4.2
Format settings, CABAC : Yes
Format settings, ReFrames : 2 frames
Codec ID : avc1
Codec ID/Info : Advanced Video Coding
Duration : 21s 522ms
Bit rate : 65.8 Mbps
Width : 1 920 pixels
Height : 1 080 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 59.940 fps
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Bits/(Pixel*Frame) : 0.529
Stream size : 169 MiB (98%)

Comments

Chienworks wrote on 12/18/2014, 1:00 PM
File size is determined by the duration and the bitrate used. Nothing else matters in the slightest. If you want the rendered file size to be 1/4 of what you're getting, use 1/4 the bitrate. Done.
Sweet Charles wrote on 12/18/2014, 1:30 PM
Dear Chienworks,

Thank you for your timely reply.

I am new to this.

Is the procedure to lower bit rate done with the filters when I'm making the Movie just before rendering?

I also checked into customizing the template.

Since I'm taking the video from my computer, is what you're meaning for me to do is re-compress the video again after Sony renders it?

Thanks Again
Chienworks wrote on 12/18/2014, 2:24 PM
When you render you can click the [Custom] button and almost all the various file type and codec templates will offer you a choice of bitrates. So, this selection happens at rendering time.

You don't want to recompress a file you've already rendered as that would compound the quality loss.
Sweet Charles wrote on 12/18/2014, 3:43 PM
Dear Chienworks,

You are exactly correct. My kbps is higher after the Sony Render than my original file.

However; when I click to make movie the only real choice I have is the XAVC S HD
1080 Long GOP 59.94 P option if I want to keep my video high quality with my resolution and
fps the same. I have looked all over for a way to change my kbps and can't find it anywhere.

Does this mean I pretty much have to go with the frame rate provided by XAVC S HD
1080 Long GOP 59.94 P which is higher than my original bit rate?

My original vbr is from about 12K to 15K kbps

Since the rendered XAVC S HD 1080 Long GOP 59.94 P kbps is a goodly margin higher I don't know if it's my eyes fooling me or not but I get a better video at a higher file size.

Thank you again for your help. I'm probably missing something.

I'd like to have my rendered video bit rate around 20 mbps but I can't seem to find how to do it.









musicvid10 wrote on 12/18/2014, 7:17 PM
Upsampling a video to a higher bitrate, framerate, or or resolution does no good whatsoever.
The quality you get in the lossy render is always less than the source, no exceptions.
Sweet Charles wrote on 12/18/2014, 8:02 PM
Dear musicvid10,

I'm referenceing very fast motion - Hawks in full flight mode.

I upsampled a good clean hd 1920 x 1080 fps 59.94 P to low range 4k, 3840 x 2160 - 1.778 : 1

bitrate = 149,807 kbps fps 59.0

In return I was able to derive and make excellent pictures from the upsampled video.

Some of these images were't even there on the original only HD.

The pictures of the birds had better color.

You can do lot more digitial with a 4k image compared to a 2k

In 4k the video is slower and gives a different perspective to the viewer more like 3D

The bark on the tree's even gets magnified so you can see better detail like a miniature bug between the crevices.

It was hard to get a software video viewer which could do it.

My file size grew to over 2,000,000 bytes from like 100k.

I am concentrating a lot on compression these days.

musicvid10 wrote on 12/18/2014, 11:29 PM
Nope, sorry. That's the kind of wishfullness that comes from reading internet blogs and some newfound theories of video alchemy circulating the social media.

We see what we want to see, and will do almost anything to convince others our point of view is true.

The only way to rule out unsupported speculation is to run controlled tests whose results can be replicated by others. Nothing else cuts it. Can we see your test results?

We can do lots of things to video to alter its appearance, including those already mentioned in this thread. None of them will "derive" (your word) additional information from the source that was not already there. It's math.

Welcome to the forums.






Sweet Charles wrote on 12/19/2014, 8:46 AM
Dear musicvid10,

I worked Sony, Adobe, Grassvalley, and Camtasia. With Camtasia I was able to get the best
1080 P rendering with lowest file size; however I was unable to upscale because 1080 P is the most Camtasia can put out.

Sony and Adobe outperformed Grassvalley taking my 1080 P fps 59.94 to 4k.

Sony had a little better overall quality, speed, ease over Adobe although both of them were able to achieve 4k.

Subject I tested was a male bluebird flying fast from its house. I used the camera snapshot function with the Sony software and was able to get the bird from standing to pushing off
and starting its flight to full flight frame by frame. I had to double click twice with each frame the first one being blurry and the second one being clear to click my shot. In my slideshow as I mentioned in my previous post I was able to freeze images in 4k which I couldn't possibly get in 2k. i.e since the bird was so fast I was able to get the bird every time it moved its wings so I had the bird in way more positions than the 2k could grab.

I was also able to get the video to run well with good sound and get a totally different perspective in 4k.

For me I like the way higher video bit rate, color, and far more detail taking the HD 1080 P
to 4k rather than in its native format.

Here are my results after upscaling from HD1080 P to 4k

[Video Information]
Codec:AVC1 - Built-in FFmpeg Decoder(h264)
Input type:AVC1(24 bits)
Input size:3840 x 2160(1.78:1)
Output type:YUY2(16 bits)
Output size:3840 x 2160(1.78:1)
Frame rate:59.94
BitRate: 149,807

[Audio Information]
Codec:twos(0x736f7774) - PCM
Sample rate:48000 -> 48000 samples/sec
Bits per sample:16 -> 16 bits/sample
Channels:2 -> 2 channels
Bitrate: 1536 kbps

General
Format : XAVC
Codec ID : XAVC
File size : 1.97 GiB
Duration : 1mn 51s
Overall bit rate : 151 Mbps
Encoded date : UTC 2014-10-17 01:58:59
Tagged date : UTC 2014-10-17 01:58:59

Video
ID : 2
Format : AVC1
Format/Info : Advanced Video Codec
Format profile : High@L5.2
Format settings, CABAC : Yes
Format settings, ReFrames : 2 frames
Codec ID : avc1
Codec ID/Info : Advanced Video Coding
Duration : 1mn 51s
Bit rate : 150 Mbps
Width : 3 840 pixels
Height : 2 160 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 59.940 fps
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Bits/(Pixel*Frame) : 0.301
Stream size : 1.95 GiB (99%)
Language : English
Encoded date : UTC 2014-10-17 01:58:59
Tagged date : UTC 2014-10-17 01:58:59

Audio
ID : 1
Format : PCM
Format settings, Endianness : Big
Format settings, Sign : Signed
Codec ID : twos
Duration : 1mn 51s
Bit rate mode : Constant
Bit rate : 1 536 Kbps
Channel(s) : 2 channels
Sampling rate : 48.0 KHz
Bit depth : 16 bits
Stream size : 20.4 MiB (1%)
Language : English
Encoded date : UTC 2014-10-17 01:58:59
Tagged date : UTC 2014-10-17 01:58:59










musicvid10 wrote on 12/19/2014, 11:15 AM
You are welcome to your opinion, but no manner of upscaling will improve on the quality of native resolution. It's just math, not fodder for speculation on anyone's part.

In fact, your choice of software upscaling in Vegas is one of the worst possible methods, assuming some actual need for upscaling exists, which is unlikely. Far better to let your hardware player or teevee upscale to 4k, since that introduces the least amount of measurable loss during delivery. This has been confirmed over and over again through controlled SSIM/PSNR testing conducted by professionals on this forum, and elsewhere.

Unlike several public gaming/hobbyist forums, we actually quantify our result here before posting. And your test results are where again?


Sweet Charles wrote on 12/19/2014, 9:15 PM
Dear Musicvid10,

I am a software guy. Once it leaves my camcorder and goes into my CPU,
I'm all SW with no Hardware.

I use a sw video player with some tweaks which can view 4k.

Thank you for welcoming me to the forums a couple posts ago or so.

I suppose you and I may have some banter about the word, "Quality."

There are an enormous indescribable amount of definitions to put into words to talk to someone about Quality just for what one see's, views.

musicvid, in your last line, last sentence within this thread your own words, "And your test results are where again?

What do you mean?

I will accommodate your request to send you my test results, but I don't think I know how
to do it. Could you give me instructions. I know I should've read Stickys.

Have a happy holiday season.

Sincerely,
Sweet Charles





musicvid10 wrote on 12/19/2014, 11:00 PM
nvm
You are judging quality by your own subjective responses, which are 100% unique to you. Projecting your personal responses onto others seems a bit, well . . . narcissistic.

I generally try to share only objectified results, which can be measured, shared, and compared by others using the same tools. Their are Quality metrics used in video production, and they've been around for some time. Perceived quality is a minor player in this arena, used mainly to find thresholds where most people can actually see image degradation consistently, long after the actual differences have been measured and enumerated.

There is no metric for quality "improvement" since no such thing exists -- the source is the reference! That said, you are entirely welcome to say "different is better," and some people may actually agree with you, albeit for their own subjective reasons.

Taken in that context, "quality" is entirely the wrong word to use. In your most recent example; the correct adjective is "bigger." The scientific question then becomes, "Show me the numbers." I'm reasonably certain you've never thought about any of this before.

Once you begin to understand the term "quantify" in the scientific sense, post back and we'll continue the discussion. In the meantime, steadfastly refuse to believe anything you read on the internet that seems too good to be true.

[b]TANSTAAFL[/B]

Best.

musicvid10 wrote on 12/20/2014, 9:47 AM
Here is a graphical representation of bitrate vs. quality in lossy encoding, which is actually the original question you posed. The representation is created by the formula Y=1-(e^-x), where [x>0]. It should be immediately apparent that source parity or 100% quality (y=1.0) can be approached, but never quite achieved. Throwing upscaling into the mix would lower the extrapolated value of Y measurably, without any physical exceptions possible. Again, I'm sure this is all brand new to you, so be patient with yourself, and we'll try to do the same.

Sweet Charles wrote on 12/22/2014, 7:58 PM
Dear musicvid10,

Thank you for taking the time to reply.

I'm going to take your advice and try to be more patient with myself and you all. You used,
"We'll try to do the same in your dialogue to me."

Musicvid10, you've told me who you think I am.

May I be a bit progressive? I think you are all about Clarity. As you see it.

musicvid10, aren't we usually an algorithm away?

I'm in 'patient mode for a bit. 'Lookin at your side

Take Good Care,

Sincerely,
Sweet Charles
musicvid10 wrote on 12/22/2014, 9:15 PM
Heh. And if you feel like being fanciful, feel free to do so.
Perhaps all that creativity will result in some video you would share here?

Factual inquiries also welcomed. Best for the holidays.



Sweet Charles wrote on 12/29/2014, 7:57 PM
Dear musicvid10,

You are A # 1.

Start and get to the truest form. I have learned from you.

And it takes work. It's definitely worth it.

You are correct.

This is my starting point.

Sincerely,
Sweet Charles
musicvid10 wrote on 12/29/2014, 10:15 PM
And looking at your handle, you have good taste in music. That was some good R&B.
Best.