Why is 76-100% quality offered if it's not recommended?

  • 1
  • Question
  • Updated 2 months ago
I'm just asking this out of curiosity... When rendering a video, if you go above 75% quality, you get the message, "Setting quality above 75% often results in a much larger file with little improvement in visual quality." So why is there even an option to go higher than 75% (or rather, why not just call the 75% mark 100%)? Is there a scenario in which the visual quality is likely to improve enough that it's worth the huge file size? 
Photo of jmgarroway

jmgarroway

  • 16 Posts
  • 6 Reply Likes

Posted 3 months ago

  • 1
Photo of Rick Stone

Rick Stone

  • 4492 Posts
  • 2002 Reply Likes
Hi there. I'll be happy to hazard a guess.

I'd be willing to bet that for the vast majority of folks, the visual content they produce with Camtasia is going to be consumed via the web. YouTube, Vimeo, ScreenCast.com, etc. And for those folks, you want to do all you can do to limit the file size in order to keep things efficient.

Now there are the random folks like me. One of the big uses of Camtasia for me is to create content that never sees the light of day on any web server. In my setup, I often produce visual content that is presented as a backdrop during live orchestra performances. And in those cases, file size simply doesn't matter and I want all the visual clarity I can get, because the projection systems used often have a tendency to "muddy up" the image quality. So if I have content that is of marginal quality, it won't look as sharp as it may with a higher quality. And since the video is being presented directly from my hard drive, size doesn't matter at all.

It will be interesting to see what others chime in with.

Cheers... Rick :)
Photo of Joe Morgan

Joe Morgan

  • 5669 Posts
  • 2938 Reply Likes
Well, when you’re uploading to sites like YouTube or Vimeo. File size does matter.

Oddly enough, 75 percent quality level videos are miles away from being good enough for either website. Case in point.

I took a 1 minute. trec recording. I rendered that video at 75 percent quality with the audio set to 320 kbps. Then I rendered the same video at 100 percent quality. I chose 320 audio because that’s the recommended setting for Vimeo. YouTube actually recommends 384.

At those levels, Camtasia produced videos with the following data rates.


As you can see at 75 percent quality. The total data rate is just under 3200 kbps. At 100 percent quality the data rate is just over 9000 kbps.

The following is an image of the recommended upload data rates for YouTube and a 1920 by 1080 video.



As you can see. YouTube recommends between 8000 and 12,000 kbps. If you are manually producing a video. Camtasia barely meets the standards at 100 percent quality. At 75 percent quality, the video doesn’t begin to touch the thresholds that it needs to be at. Had I set the audio to 384 kbps. The bit rate would’ve been somewhat higher.

Now let’s take a peek at Vimeo.



This screenshot is settings you would use for Premier Pro cc. There are more settings available for Premier Pro so I highlighted the ones that would apply to Camtasia. Vimeo recommends 10 to 20,000 kbps. At 100 percent quality, Camtasia does not even meet that threshold.

So you see when it comes right down to it, Camtasia fall short at 100 percent quality.

Now let’s take a peek at Premier Pro and what happens to the same footage when producing for Vimeo and YouTube.


If you look at the name of the title in the properties. You can see the default output I selected in Premier Pro. The Vimeo and YouTube outputs were nearly identical at 11,050 kbps. This bit rate falls within the guidelines of both websites. With the exception of the audio. YouTube recommends 384 kbps.

For the heck of it I rendered one as a high quality MP4. That one rendered over 12,000 kbps. So Camtasia falls short of HDTV quality. According to Adobe's guidelines. I'll take their opinion over TechSmiths. 

The H264 format is the H264 format. You can take a video and run it through handbrake and overly compressed the same video and have it look good for viewing purposes. The file size will drop as a result of that compression. So you might think to yourself well, file size doesn’t matter. It looks "Good" Not fantastic. Plus, if you upload that overly compressed file to YouTube. It won’t process nearly as well is the original file would have.

If you take that overly compressed handbrake video and stick it in Premier Pro. Try to apply color corrections so forth and so on. It won’t work well. You will find that the original video will accept changes much better than the handbrake version.

The handbrake version won't re-render well either. Lower quality is lower quality. You can't reduce file size without reducing quality.If you think you can, think again. 

Large files are large for a reason. Red Cameras shoot uncompressed video at up to 8K for superior editing capabilities. The file sizes are staggering.

100 percent quality in Camtasia? It can get better than that. Why Camtasia chooses to call it 100 percent quality? Well, it’s the highest setting Camtasia has to offer.

Regards, Joe


(Edited)
Photo of rg

rg

  • 105 Posts
  • 56 Reply Likes
Thanks for the excellent explanation.
Photo of Dave O'Rourke

Dave O'Rourke, Senior Software Engineer

  • 1419 Posts
  • 376 Reply Likes
Joe, the thing you're missing is that screen video is *much* more compressible than camera video.  The guidelines on YouTube, Vimeo, and Adobe apply to camera video, but they're far too high for screen video.  Since YouTube and Vimeo transcode your video, what matters is how clean they look visually, not the file size.  See my post below for a more detailed explanation.

You are correct that you can't get the quality back once it's gone.  YouTube and Vimeo *need* you to upload a high quality source for that reason.  They can't fix the quality once it's gone.
Photo of rg

rg

  • 105 Posts
  • 56 Reply Likes
I think it's a bit like Spinal Tap -- "these go to eleven."  Well, Camtasia goes to 100!

Seriously, I find stuff like "percent quality" less useful than providing data rate.  It would be more helpful to offer a slider or radio buttons for various data rates, with pop-ups that roughly explain/describe quality levels.
(Edited)
Photo of Rick Stone

Rick Stone

  • 4492 Posts
  • 2002 Reply Likes
LOL, LOVE IT!
Photo of pearsonweb

pearsonweb

  • 8 Posts
  • 9 Reply Likes
Indeed - every time a upload one of my corp training videos rendered in Camtasia to vimeo, it says "Your video could look much better if you gave it the proper resolution" or something like that. In my experience most of my training videos are consumed via mobile device or PC. But we do have occasion where the video is displayed on a 60" HDTV or projected - that's when I feel the 'missing' pixels or compression is noticed. I thought the rule of thumb was "produce at the highest quality/setting possible and let the broadcast device or stream throttle it as necessary" but maybe that's not quite accurate. My coworkers seem obsessed with getting the smallest file size possible but I take the opposite stance. They set their audio to mono instead of stereo and compress the audio to 96kbps and insist the user can't tell the difference between  a 320KB stereo audio and 96KB mono. I've not done official testing but I've always been a quality kind of guy - bandwidth and storage are (relatively) cheap - why not go for the biggest/best production you can get? 
Photo of david

david

  • 16 Posts
  • 11 Reply Likes

I agree pearsonweb..  Why slight the audience with a n OK presentation, if a great presentation, including audio, is available?
Photo of rg

rg

  • 101 Posts
  • 52 Reply Likes
I agree.

In the old days of the web, the recipient's bandwidth was a primary determinant, so if you had a massive stream of video/audio it would cause hiccups while the recipient's computer downloaded more data chunks.  Sometimes a high-rate stream would cause a freeze.  That was why people opted for smaller file sizes.  For playback from a disc or hard drive, a constant (not variable) bit rate was crucial, and on many PCs a rate of 4000 was sufficient for, uhm, reasonable quality.

All of this has evolved.  Now, communication between the browser and the provider can adjust and optimize the data stream.  Screens have higher resolutions and come in ever larger sizes.  Quality can once again be a goal, without sacrificing the ability to play video without freezes.  The obsessions with reducing file size and data rate are no longer requirements (although for most applications, I'd still question whether stereo typically makes any difference -- this ain't The Doors' Soft Parade after all).  While there's an audible difference between 320kbps and 96kbps, most people would not hear a difference between 320 and 192.  Most cannot hear a difference between audio at 44.1 vs 48 sampling rates unless you do an A/B comparison.

Even with all of the advances, if the target audience will be viewing on their phones, I wonder if most people would notice (or care about) the difference between a 3k versus an 11k data stream.

Side note:  Many podcasts stream/download MP3 files with 96kbps to help ensure stability.  The sound is crunchy, but the subscribers keep listening.
(Edited)
Photo of Joe Morgan

Joe Morgan

  • 5598 Posts
  • 2910 Reply Likes
Perfect example pearsonweb
Photo of jmgarroway

jmgarroway

  • 16 Posts
  • 6 Reply Likes
Thanks everyone! I must admit that the responses have made me realize that my knowledge of video production is actually quite shallow (I've dabbled a tiny bit in After Effects, but otherwise, I've always worked with Camtasia and used mostly default settings). I really appreciate the detailed responses :-)
Photo of Dave O'Rourke

Dave O'Rourke, Senior Software Engineer

  • 1419 Posts
  • 376 Reply Likes

Sorry, I'm late to this thread.  Bitrate and encoding is a complex topic.  I'll do my best to shed some light on what's going on here.

To answer the question from the OP, the warning is there because we had a lot of people who would push the quality all the way up to 90% or 100% and then complain that they got a much larger file with little improvement in visual quality.  That's expected, because the quality slider is hooked up (inversely) to a constant quantizer parameter. The net effect is that linear increases in the quality % result in an exponential increase in file size.

So why not just remove the upper quality range as an option?  Well, as pointed out in this thread, sometimes users *want* max quality, and don't care about file size.  There are times when that's useful. But note that the visual quality difference between say 85% and 100% is sometimes very slight, yet the file may be *much* bigger.  A larger MP4 file can cause long upload or download times, and in some cases, playback stuttering if the MP4 is placed on a network drive or other drive with slower disk access times.

Now let's talk about YouTube and Vimeo.  When you upload a video to these sites, they are transcoding your video (making multiple copies of your original) at different dimensions and bitrates.  This happens during the "processing" step, which begins immediately after the upload completes. This is done so that they can optimize the viewing experience for many different devices, bandwidths, etc.  So the video you're watching on YouTube isn't actually the same bits you uploaded. Knowing that fact changes things a bit. Since the video coming out of Camtasia isn't the video being downloaded by your viewer, the file size only affects your upload time to YouTube and Vimeo, and this has no effect on the experience of the viewer, because these sites will fix that for you.  So for YouTube and Vimeo, the video and audio quality are the most important things.  The encoding guidelines are given to help you upload videos of high quality.

Ok, onto bitrate.  The bitrate setting on an encoder is a constraint.  It sets a target or max amount of bits (file size) that the encoder is allowed to use for a second of video.  If a section of video hits this cap, the encoder lowers the visual quality for that section in order to meet the bitrate constraint.  If the bitrate is not allowed to vary, and the bitrate is too low, the video may have complex, high motion sections that are blocky or blurry.  Starting from that video, you can't get that quality back. You have to re-produce starting from the original at a higher bitrate so that those sections don't hit the bitrate constraint.  That's why YouTube and Vimeo list bitrates that are on the high side. They don't want you to upload video that's been over compressed, because they can't do anything to fix it.

But there's another way to encode video.  It's called constant quality variable bitrate encoding.  The idea here is to NOT put a bitrate constraint on the encoder.  Instead, we allow the bitrate to vary, as needed, according to the complexity of the video.  This has the desirable effect that the video maintains a similar visual quality throughout, while only using as much file size as is necessary for a particular section of video.  The result is a well compressed video that doesn't get blocky or blurry during any high motion sequences.

This second approach is Camtasia's default for MP4.  For complex sections of the video, it will use a higher bitrate than for less complex sections.  So the content of the video affects the file size.  This is where screen recordings are different than camera footage.  Consider a recording of your screen at 30 frames per sec. Most of the time, the screen is very static and doesn't change much from one frame to the next.  There's a lot of redundant temporal information, and that makes the video highly compressible. The same is not true with real world video that you record with your phone or video camera, especially if there's motion from the camera, or from whatever is being filmed.  This content is less compressible, and generally requires a larger size to maintain the same visual quality.

So why does Vimeo sometimes warn when you upload a video from Camtasia?  Put simply, because the video is smaller than they expect. They look at the file size of the video, and the length, and if it's smaller than a threshold, they guess that the video may be of low visual quality.  But is it? Watch it and decide for yourself. The important thing is that you're uploading something of high visual quality.  Well compressed camera video is going to be much larger than well compressed screen video. Their threshold probably works well for camera video, but not to low-motion screen video.

Sorry for the long post.  As I said in the beginning, bitrate is a complicated topic.  If you've read this far, I hope you've found this post helpful.

Photo of jmgarroway

jmgarroway

  • 16 Posts
  • 6 Reply Likes
I did find it helpful! Thank you very much :-)
Photo of Joe Morgan

Joe Morgan

  • 5669 Posts
  • 2938 Reply Likes
Well Dave O'Rourke, this is in response to your comment about screen recordings being much more compressible than camera video.

That depends entirely on what the screen recording is.
Crysis 3. A video game I happen to own that was released in 2007. Was so demanding to run there wasn't a consumer computer available that could run it at its highest settings.

Granted, that has nothing to do with the actual recorded image quality.

However, with smoke, mist, haze, clouds and explosions. Sunlight and shadows. City buildings in ruins.  Coupled with particle effects and Helicopters flying by, mayhem all around. Crank up the 4K at up to 60fps to boot.

Heck, even camera video compresses more than that. I’m just yanking your chain a bit here. Not trying to offend here, just kidding around.

But yeah, I understand compression. And that a static screen, especially without much color can lead to smaller files.

I probably should have run more tests before posting what I did above. But TechSmith’s rendering results/You Tube settings don’t make any sense. Not compared to Adobes. Not in my limited tests.

So above, the 75% quality video I created in Camtasia.And the You Tube Quality one I created in Premiere Pro video.

I took Identical frames from each. Stacked them in Photoshop. Zoomed in and panned around. Reducing the opacity of the frame above off and on. They were nearly identical. The differences were so minor; I could have ignored them. But I saw slight differences in red. That was it. I just wanted to cover that first.

So then I got curious about Camtasia’s You Tube uploads. Uploading through Camtasia directly.

 I have DSL. It took me well over 20 minutes to upload a 1-minute video. I knew something was seriously off with this picture. It only took me about 5 minutes to upload the Premier Pro video.

So I did some work outside and thought about it. Came in and opened the Windows temp folder. Rendered the same 1-minute video to You Tube. Figured I could see how big that temp file was while it uploaded. It rendered but then wouldn’t connect to You Tube. Some error occurred so it offered to save the file locally. Awesome/Perfect! I saved it.

 It was 335mb at just under 47,000kbps. WTH? That explains the 20+ min upload.



Premier Pro saved the same at a more reasonable 117mb 16,400 kbps. This is Crisis 2 by the way. I haven’t had Crisis 3 installed in a while. I may have to this weekend, now that it’s on my mind.


Both uploads look and play the same to me. Here’s the Premiere Pro version. You'll probably have to watch them on you tube to see them in full screen and full HD.



And here’s the Camtasia version.


It would be very helpful if Camtasia had Estimated file size before rendering. Premiere Pro tells you the approximate file size of a video before you render it. You would have to do the math to determine approximate bit rates, but at least you have some idea of what’s going on.

Anyway, enjoy the weekend. Regards,Joe







(Edited)