Make Camtasia Studio 8 Rendering Faster!

  • 13
  • Idea
  • Updated 4 years ago
  • Implemented
Multi-Core Support in Camtasia Studio 8 - It appears that the rendering program for MP4's in Camtasia 8 isn't able to render using more than 1 core. Make it take full advantage of today's hardware with multi-core video rendering.
Photo of Justin Gehring

Justin Gehring

  • 9 Posts
  • 1 Reply Like
  • Hopeful

Posted 7 years ago

  • 13
Photo of Randy Schott

Randy Schott

  • 396 Posts
  • 47 Reply Likes
Hi justin,

We do actually use multiple cores in Camtasia Studio. We don't always use every last bit of them, but in some areas we try to do it:

-Our third party H.264 decoder spreads the work across threads
-We use separate threads for rendering audio/video
-We use DirectShow under the hood for decoding videos, it makes extensive use of threading
-We buffer the next frame of decoded video for each file in the background while we are composing the current set.

I thought the encoders we use for MP4 had support for multi-thread encoding, but maybe it doesn't. It's not usually the bottleneck in exporting (that is usually generating the frames to be encoded). But I guess if you have a simple timeline, that changes things a bit.

We *are* investigating using the GPU to accelerate encoding of files. Our vendor already supports it, just a matter of seeing if we can make it work with Camtasia.

Thanks for the great feedback. Hope you're enjoying CS8. :-)

Randy
Photo of Justin Gehring

Justin Gehring

  • 9 Posts
  • 1 Reply Like
Hey Randy,

It's interesting watching the performance from version to version. In Camtasia 7, with the normal mp4 (web) encoding setting, 2 of my 4 cores would max out, while the other 2 sat mostly idle. My machine would render a 1 hour video in about 5 minutes (give or take), and task manager would report my processor at 50%ish.

Now with Camtasia 8, it seems my processing time has almost doubled to 10 minutes, and my processor is capping at about 20 to 25%... So I figured either the new encoder isn't setup to use all the cores, or something else is funny. I tweaked my encoding settings a little bit, but still didn't get much more gain out of it.

I am using a very simple timeline (1 video, 1 audio, 1 pan and zoom at the beginning, no other edits). Basically it's a record, scale, and export.

I have a hard time seeing my hard drive or any other piece of hardware being the bottleneck, as I am on a 800MB/s SSD raid for read, and a 600MB/s SSD (these are 2 seperate drives) for write. (I'm a little bit of a performance junky :-)).

I assumed this was a new encoder in Camtasia 8, as it looks like it did a slightly better job compressing the file size (50mb file vs. a 40mb file). That may have also been my render settings though, haven't done a true 1 on1 comparison..

All I know is that out of the box, Camtasia 7 is feeling faster rendering wise than Camtasia 8.

But thank you for the great work and software!
Photo of Andyboy

Andyboy

  • 4 Posts
  • 0 Reply Likes
Randy:

Looking  to drive two 4K displays, I am considering an nVidia Quadro GPU or EVGA GTX 780 -- or maybe running two of them using SLI along with  a six-core processor 4930K CPU.

Does Camtasia now use multiple cores on a CPU--multithreading- any better than when you responded to this post in 2012?
.
Also, Camtasia is not on nVidia's list of video editing applications that are optimized to run on the Quadro family of GPUs. Will that change anytime soon? 

I'm asking because I don't whether to buy two 780s or two Quadros.

Thanks
Andy

 

Photo of Joe Morgan

Joe Morgan

  • 7116 Posts
  • 3865 Reply Likes
The Camtasia editor can only handle 2048 X 2048 Max Resolution.
If you want to work in 4K. You will need a different program for editing. 

Regards, Joe

(Edited)
Photo of Andyboy

Andyboy

  • 4 Posts
  • 0 Reply Likes
That's must be planned for a new version. No? TechSmith will need to keep up before Premier eats its lunch.  
Photo of Randy Schott

Randy Schott

  • 396 Posts
  • 47 Reply Likes
Interesting....

Testing internally, we mostly saw that CS8 outperformed CS7. But again, that was with the frame generation. The encoders are actually identical to the ones in CS7, we just tweaked the default settings. There are a handful of things that could be affecting the performance. Ultimately, we just need to do a better job with the multi-core optimization. I know we will be constantly exploring ways to better leverage multi-core machines, since processors will be scaling in number going forward, not in clock speed.

It's something I'm personally very interested in, but it's VERY difficult to do it correctly :-).

One thought is that if you are scaling at production time (production dimensions are not equal to editing dimensions), I think the scaling algorithm changed in CS8 to something that is a little slower (but provides better results).

Also, I'm assuming you aren't on XP. But if you are, performance might be worse in some situations, particularly production.

I'm hope you're enjoying CS8 otherwise though. Hopefully we can get it performing to your expectations.

Randy
Photo of Nicole Simon

Nicole Simon

  • 24 Posts
  • 22 Reply Likes
Randy do you know a way to tell windows to always run camtasia on high priority instead of having to do that manually every single time we start the program? thanks
Photo of demarest

demarest

  • 99 Posts
  • 10 Reply Likes
I just did a quick Google search of "windows shortcut to always run a program in high priority" to see if it was possible and it looks like it is. This is the sort of thing that users should be doing, not program providers or else all programs would be made that way and the control of such would be meaningless.
Photo of Kevin Liu

Kevin Liu, Staff Software Engineer

  • 113 Posts
  • 21 Reply Likes
Hi Justin,

Just wanted to check if you are running in GPU accerated mode. When the timeline is in pause mode (not playing in preview), select a video clip on timeline, then see if you see a blue dot in the preview. If yes, then it is in software rendering mode (a fall back when GPU has problem to render the timeline preview or production). If you see a white circle, it then is in GPU rendering mode. Also you can switch between these in Tools->Options->Use GPU acceleration if available (you need to reload the project for this switch change to take effect). Basically, software rendering mode should be very close to what CS 7 does if not better, or there might be other causes we want to find out.

Thanks,

Kevin
Photo of Jeff E.

Jeff E.

  • 9 Posts
  • 1 Reply Like
I'm a HUGE Techsmith fan and love the new features in CS v8, but I am also noticing a very significant increase in the time it takes (versus version 7) to render videos in both the "Share to Screencast.com" setting and when using the custom production settings to export as MP4.

When Sharing To Screencast.com, I tried to make some adjustments to the settings, but it still takes much longer to render (3-6 times as long, I'm estimating).

Also, I quite frequently create 1366x768 screen capture videos, and the smaller/default viewing size on Screencast.com always looks great. After the upgrades however, the smaller viewing size on Screencast.com has a significant decrease in clarity, especially on text.

I'm guessing and hoping that there are just some setting that I need to optimize?
Any thoughts would be appreciated.
Photo of Jeff E.

Jeff E.

  • 9 Posts
  • 1 Reply Like
Side note: I'm a bit baffled by this because: The newest version of SnagIt's video capture renders so fast that I hardly notice it's started! And, the default settings when uploading to Screencast.com from SnagIt video capture look crystal clear, in the default smaller viewing size, not just the 'full size' view on Screencast.
Photo of Dave O'Rourke

Dave O'Rourke, Senior Software Engineer

  • 1443 Posts
  • 412 Reply Likes
Hi Jeff,

Snagit's video capture actually encodes to MP4 on the fly, at 10fps or less, dropping frames if the video encoding can't keep up. So when you stop recording in Snagit, you have a video that's ready to share. By contrast, Camtasia Studio 8 records at 30fps as source footage for the editing/production process. When you share to Screencast in Snagit, you're just uploading the video that you recorded. When you produce for Screencast in Camtasia Studio, we first produce your timeline to MP4, and then upload to Screencast.

As for the visual clarity on Screencast.com, if you're capturing the same content at the same dimensions, I'd expect the see the same result on Screencast.com from both Snagit 11 and Camtasia Studio 8. If you could post a screenshot of each (Snagit > share to Screencast > paste links here) we could investigate further.

Thanks for the feedback.
Photo of Jeff E.

Jeff E.

  • 9 Posts
  • 1 Reply Like
Thanks for the info dro! I'm going to paste a link to a recent test at the bottom of this discussion. I really appreciate you guys working so diligently (and late into the night) on this! This is what sets you guys apart from your competition..your incredible commitment! Keep up the good work :)
Photo of Justin Gehring

Justin Gehring

  • 9 Posts
  • 1 Reply Like
I am in GPU accelerated mode. I see the white dot, not the blue. I tried shutting it off, just to see what happens, and things did get slower. I did up a quick video here showing a the best equal comparison i could come up with (although to be honest, I think CS7 actually has the better (higher quality render) settings... i probably could have eeke'd out a few more seconds on the CS7 side using my other render setting...

Regardless, here it is:
http://justingehring.com/CAM/SpeedTes...

CS7 beat CS8 by effectively 2 minutes for this particular render (8 seconds to 5% = 160 seconds), and I'm 95% certain it's not my camrec settings (I used the same camrec), or my render output settings (although i can get some gains with some tweaks, on either one, it seems across the board 7 keeps coming out ahead).

Hope it helps with testing, and I'll add this, as long as there are CPU cycles not being used, there's some performance to be had :-), good luck!
Photo of ms

ms

  • 162 Posts
  • 22 Reply Likes
Running v7 against v8 - v8 is considerably slower on 2 systems here.
Photo of JK Swopes

JK Swopes

  • 18 Posts
  • 4 Reply Likes
I also immediately noticed a major increase in rendering time with v8 vs v7 hoping this is addressed soon.
Photo of Kevin Liu

Kevin Liu, Staff Software Engineer

  • 113 Posts
  • 21 Reply Likes
Hello,

Some update on this. I tested a simple project on my machines (dual-core and quad-core), in both cases, 8 were faster than 7 in rendering. And I also noticed all the cores were busy at the time which was expected. So I was wondering if you could try a few things and see we can figure out the root cause:

1) I assume you are on a 64-bit machine which system memory usage isn't an issue (note CS 8 may need more system memory than CS 7). If you are running a 32-bit machine, go to the very end of this post, follow the instruction there to turn on the 3GB flag to take advantage of the memory you have for a 32-bit app (eg: CS 8).

2) Check to make sure you have not turned off any of your cpu cores:
a) In search box or start->run, type in msconfig, then run it;
b) In the System Configuration window click on the Boot tab and click on Advanced Options button; in the pop up window, make sure the "Number of processors" option is NOT checked; If it is checked, uncheck it and reboot your system;

3) Update the DirectX run time to the latest:
http://www.microsoft.com/en-us/downlo...
Reboot when this is done

4) Update windows to have the latest updates in there;

5) Try running CS 7 and 8 again with the same test project and see if it helps. To make sure you are comparing apple with apple, use MediaInfo or other tools of your choice to check on the produced MP4 and see if they have the same profile/level as well as dimensions etc, otherwise, you can tweak the settings in CS 8 to match it in 7 (custom path under video settings tab in CS 8);
6) Please do report back if you still see issues etc.

============== Turn On 3GB Flag =======================
1) Note You Don't Need to Do This If You Have a 64-Bit OS on Your Machine
2) Turning on the 3GB flag has certain risk especially on a 32-bit XP, it is on your own to do this !!!
====================================================
How to enable 3GB switch on your OS?

Enable the 3GB switch on 32-bit Windows XP
• Right-click My Computer. Click Properties.
• In the System Properties dialog box, click the Advanced tab.
• On the Advanced tab, under Startup and Recovery, click Settings.
• In the Startup and Recovery dialog box, under System startup, click Edit. The Windows boot.ini file will be opened in Microsoft® Notepad.
• Create a backup copy of the boot.ini file. Note: Boot.ini files may vary from computer to computer.
Select the following line in the boot.ini file:
multi(0)disk(0)rdisk(0)partition(2)\WINDOWS="Microsoft Windows XP Professional" /fastdetect
•Press Ctrl+C to copy the line and then press Ctrl+V to paste it immediately below the original line.
Note: Your text string may be different from the text string in this solution, so be sure to copy the text string from your boot.ini file, and not the text string included here.
Modify the copied line to include “ /3GB”, as shown in the following example:
multi(0)disk(0)rdisk(0)partition(2)\WINDOWS="Microsoft Windows XP Professional 3GB" /3GB /fastdetect
Note: Do not overwrite any existing lines.
•Save and close the boot.ini file.
•Click OK to close each dialog box.
•Restart your computer.
•During startup, select the 3GB option. If you do not select the 3GB option, the system will default to the 2GB total memory setting.
Note: If problems occur during startup, you may need to update some of your drivers.

Enable the 3GB switch on 32-bit Windows Vista/7
•Right-click Command Prompt in the Accessories program group of the Start menu. Click Run as Administrator.
•At the command prompt, enter "bcdedit /set IncreaseUserVa 3072"
•Restart the computer.

Disable the 3GB switch on 32-bit Windows Vista/7 (Do this only if you want to remove the 3GB flag)
•Right-click on Command Prompt in the Accessories program group of the Start menu. Click Run as Administrator.
•At the command prompt, enter "bcdedit /deletevalue IncreaseUserVa"
•Restart the computer.

For more information on the 3GB switch, refer to the following Microsoft MSDN article:
http://msdn2.microsoft.com/en-us/libr...

Thanks,

Kevin
Photo of Kevin Liu

Kevin Liu, Staff Software Engineer

  • 113 Posts
  • 21 Reply Likes
I have tested a few other projects for 7 and 8. These projects have video, audio, captions, callouts, markers etc combination and 8 always faster in term of rendering speed - 30 to 50% faster by average. The bigger dimensions and more tracks, GPU weights more, hence faster. Your mileage may vary, but I will be surprised if the result is opposite. So I'd be interested in getting some of your test projects if possible.

As a side note, in addition to updating your DirectX runtime and Windows system, you may also want to update your video card driver etc.

Also to be fair for CS 7, you may also want to use the memory patch found here: http://techsmith.custhelp.com/app/ans...

The steps of the 3GB flag for 32-bit OS and App to take advantage of this memory patch also apply to CS 7.

Thanks,

Kevin
Photo of Justin Gehring

Justin Gehring

  • 9 Posts
  • 1 Reply Like
Just as a note Kevin, my tests have been fully patched, etc. I haven't done a test with a ton on the timeline, but if it helps, I'd be willing to run that test for you.

Most of my tests (just so you understand the scenario that's playing out), are straight recordings (maybe with a cut at beginning and end, and a zoom/resize at the beginning). My use scenario has be making 1 hour videos and then turning them around as fast as possible. I still have yet to do one more comparison making sure everything is evenly matched... But again, I'll draw your attention to pure processor cycles in my video... while more processor power != faster (although that seems to be the case for me)... not using the full processor == not using maximum speed.

It'd be nice if there was a way during render to see where the bottleneck was. I really have a hard time believing it's on my SSD... Graphics card is possible depending on much it's being relied on... I'll see if I can pull out a performance chart there when I run my next test.
Photo of Kevin Liu

Kevin Liu, Staff Software Engineer

  • 113 Posts
  • 21 Reply Likes
Thanks Justin, I am going to try a few more long recordings by following your workflow.
Photo of Kevin Liu

Kevin Liu, Staff Software Engineer

  • 113 Posts
  • 21 Reply Likes
Minor update: I tried a few more long recordings (30 to 60 minutes long, several to 20GB each in camrec file size), and CS 8 is about 50 to 100% faster than CS 7 in rendering speed (I tested on a dual-core 32-bit Win 7 machine).
Photo of Dave O'Rourke

Dave O'Rourke, Senior Software Engineer

  • 1443 Posts
  • 412 Reply Likes
On the topic of rendering speed...

One thing to note is that Camtasia Studio 7 recorded the screen at 15fps by default. Camtasia Studio 8 records the screen at 30fps by default.

By default, the "Automatic" production frame rate will use the highest frame rate of all the clips on the timeline, with a min of 1fps, and a max of 30fps. If you go down the custom path in the production wizard, you can set the frame rate manually if you like.

So if you record + produce in CS7, you'll be producing a video at 15fps. If you record + produce in CS8, you'll be producing a video at 30fps.

To get an apples-to-apples comparison of the render speed, you'd need to do the following:
1. Use the same video on the timeline in both CS7 and CS8
2. Use the same editing dimensions in both CS7 and CS8
3. Use the same production settings (framerate, dimensions, format=MP4, etc.) in both CS7 and CS8

Of course you can drop the recording or production framerate in CS8 down to 15 if you like, to match CS7. This might speed things up a bit, but it will make your final video a bit more choppy to watch.

Hope this helps.
Photo of Justin Gehring

Justin Gehring

  • 9 Posts
  • 1 Reply Like
Interesting note... in most of my tests, I'm rendering at 1FPS and recording at 15FPS... I'll do another direct comparison tomorrow between the 2 as best I can... and try to record it for you like my earlier video.
Photo of Jeff E.

Jeff E.

  • 9 Posts
  • 1 Reply Like
I'm not sure what the default audio kbps is in v7 for the Share to Screencast output, but in this test I set v8 to 56kbps. I must be doing something wrong. This test uses the same camrec file in both versions. Version 7 renders in approx 4 minutes, and version 8 renders in about 25 minutes.

I'm almost certain that the quality of the default size in Screencast was noticeable better as well, although during my test (as you will see from the video), v7 uploads fine, but v8 tells me that my Screencast plugin is out of date, so I couldn't do the side-by-side Screencast.com comparison I was hoping for in this recorded test.

http://screencast.com/t/BCuCPRhkTAM

*another side note: After uploading to Screencast from SnagIt, occasionally (as in this case) the index time in the bottom player bar shows the video only being 6:50, yet it's actually about 35 min long, and will just continue to run after the player bar gets to the end.
Photo of Jeff E.

Jeff E.

  • 9 Posts
  • 1 Reply Like
Btw...not complaining here, just trying to help get my issues figured out ;)
Photo of Glenn Hoeppner

Glenn Hoeppner, Employee

  • 1838 Posts
  • 337 Reply Likes
Jeff, the duration issue with Snagit's mp4's on Screencast.com is a known issue. There's a temporary fix until we can get the actual issue corrected. See this support doc: http://techsmith.custhelp.com/app/ans...
Photo of Jeff E.

Jeff E.

  • 9 Posts
  • 1 Reply Like
Thanks Glenn! I hadn't even searched the forum for this issue as it's not a show stopper issue for me. That being said, I'm glad it's being worked on, and thanks for the link to the temp fix!
Photo of Jeff E.

Jeff E.

  • 9 Posts
  • 1 Reply Like
I'm going to try a re-install tonight or tomorrow and will report back!
Thanks!
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
Same disappointment here...

Same project, 18 min long, produced for youtube (win 7 x64, i7 2600k, 16GB Ram, SSD, AMD Radeon HD 6970)
CS7: 1m 17sec
CS8: 2m 37sec

I've done all checks you suggested (number of processors, directX, GPU enabled ecc.).
Photo of John S. Richards

John S. Richards

  • 391 Posts
  • 60 Reply Likes
Check you have identical settings to record and produce.
CS7 may be recorded & rendered at 15fps while CS8 is at 30fps.
CS8 is faster when all settings are the same.
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
Thank you; I thought youtube preset profile was identical in CS7 / 8 (someone can confirm this?).

I don't know how to modify or show youtube profile settings: it seems it's all automatic and when you choose 'share to youtube' the step where you can review settings is skipped.

I'll try with manually configured profiles but the one which I care is youtube ...
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
I followed an alternative procedure to test settings for youtube preset profils in CS7 / 8: I disabled internet connection so after producing CS ask me for saving videos (surely there is a less rude way!). In the image you can see that the videos properties (v7 on the left) are almost identical....
Photo of John S. Richards

John S. Richards

  • 391 Posts
  • 60 Reply Likes
Sending to Screencast.com is always the best bet.
What viewers see on Screencast.com is exactly what you produced.
You maintain control over the entire viewing experience.

YouTube uses a video compression algorithm to compress any video uploaded.
So what you may see on YouTube might not be as good as on Screencast.com.

The Camtasia produce to YouTube setting produces a video most compatable with how YouTube works and manages an uploaded video. I expect CS7 and CS8 produce to YouTube settings will be similar, and most likely improved by CS8 as you are now rendering at a default of 30fps. I have not checked to see if there are any settings differences and cannot confirm.

You can of course use your own custom settings, then upload the produced MP4 video direct to YouTube by logging into your YouTube account. But still you are at the mercy of the YouTube video compression software once you have uploaded.

The real problem with YouTube is that a viewer's visual experience is typically degraded at the beginning of a video, since YouTube typically starts serving your video up in a low quality format, forcing viewers to stop, select a higher quality view setting, and re-start the viewing process all over again. This is crazy.

A second problem with YouTube comes at the end of the viewing experience. Viewers typically see a mosaic end screen which is nothing but visual noise, connecting viewers to other videos that are not yours, which may have little if anything to do with your subject matter.

YouTube = Use at your own risk?
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
Thank for the suggestions. I'll think seriously about that.

You speak about 30fps but in the image I've uploaded you can see that the two videos have the same 5fps (my standard for software training videos) and all other properties are almost identical. So I'm confused. Perhaps you're suggesting that the weird result is limited to videos produced for youtube? I'll try with user profiles with identical (as much as possible) settings.

Thank you again.
Photo of Larry Bob

Larry Bob

  • 3 Posts
  • 0 Reply Likes
You make a good point, however Screencast.com bandwidth restrictions for free accounts are WAY too low.

eg. I uploaded a 100 MB video a while ago to test out the service. This video alone would only need to be watched a total of 20 times in a month before I went over my restriction. And I planned to upload other videos too. I understand you're trying to push the premium version, but this is very offputting.

Having it upped a bit would certainly entice me to use the service.
Photo of Kevin Liu

Kevin Liu, Staff Software Engineer

  • 113 Posts
  • 21 Reply Likes
Hi Fabrizio,

I just did a quick test of this youtube path, I drop the wildlife.wmv - Windows OS provided sample video, here is my setup on CS8 (I did similarly on CS 7):



I then go directly to produce using youtube path, the rendering time comparison (first step only, the second step is uploading, I didn't use and I don't remember we changed anything for the uploading part, so I just ignored it):
CS 8 - 23 seconds
CS 7 - 44 seconds

My computer:


So would you please try the same file for both CS 7 and 8 and see how it goes?

Thanks,

Kevin
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
I tried leaving 'recording dimension' (HD) and obtained results comparable with yours:
CS8: 10 secs
CS7: 20 secs

No doubts here: CS8 rules :)

My other project was recorded in CS7, camrec format, 34 clips (total 18 mins); I'll try with a monolitic camrec of 18 minutes of same type (such as a software training video from you tube) just to investigate if clips number affects cs8 way of rendering; other suggestions?

May be that CS7 camrec format is optimized for that version and not for CS8? In other words if I had recorded the same sequences in CS8 camrec format the rendering would had be faster? (sorry if I'm misusing english verbs here ..)

Thank you for taking care,
Fabrizio
Photo of Justin Gehring

Justin Gehring

  • 9 Posts
  • 1 Reply Like
Here's my latest test.

http://www.justingehring.com/CAM/Spee...

- Same camrec (same input)
- Same Settings (shown in video)
- CS7: 1 minute 7 seconds
- CS8: 3 minutes 48 seconds.

I threw out a few stranger ideas in the recording... Hope it helps.
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
Hi, Justin. Did you record with CS8 recorder or CS7?
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
I apologize if my questions seems silly; your video is clear but I'm not a native english speaker so it's possibile I missed this bit of information in your narration. Thank you for sharing your time with these tests ;)
Photo of Justin Gehring

Justin Gehring

  • 9 Posts
  • 1 Reply Like
I'm pretty certain it was with the cam8 recorder... but i'm not 100% certain. I'll run the test again perhaps later today, with a new 5 minute recording being 100% certain that I used the cam 8 recorder. As a note though, my cam 7 recorder was set to Techsmith Codec, 10FPS, Keyframe every 80 frames (if that was the camrec i was using). My cam 8 recorder is set to the techsmith screen codec 2, still set at 10fps.
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
I've captured with CS8 recorder Justin's video, about a monolitic 300MB camrec.
Produced for youtube (so as Kevin suggests -> parameters should be almost identical)

Here the results:

CS8: 1min 58 secs
CS7: 1min 40 secs

At this point (with wmv format CS8 ruled in my test performed as Kevin suggested) I suspect the negative ingredient here could be the camrec format: may be it slows down the process more in CS8 than in CS7? I'll repeat the test recording in avi with same recording parameters...
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
Recaptured Justin video, avi format (about half the size of the camrec: 140MB); produced as usual for youtube:

CS8: 1min 35 sec
CS7: 1 min 47 sec

So I think I must do a last test: convert the avi in wmv to see if the delta in rendering time is so evident (in favour of CS8) as was with the first test suggested by Kevin.
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
First a litte update regarding previous test: I didn't realize avi capturing excludes system audio. So I exported audio from the camrec format of the same video and added in the projects; times did'n noticeably changed (a bit of suprise here).

Here instead the results using wmv format (captured with MS Expression 4 Recorder and encoded with the corresponding encoder):

CS8: 3min 41 sec
CS7: 5min 12 sec

These results are consistent with previuos. So the only thing 'clear' is that CS8 rules when wmv is involved. But this could mean simply that CS7 was not so optimized for wmv format.

Disappointment remain: CS8 still seems to me not to take strength on GPU support as expected :(

May be CS8 is better optimized for CUDA (nvidia) than for ATI stream (AMD): often I see better support for CUDA around the world ...

Justin, is your video card from AMD or Nvidia?

Kevin can you reproduce my test (at least that with camrec format)?

If some kind CUDA guy was interested in doing some test, here the link at the camrec/CS8recorded (194MB): www.camuso.it/Download/Justin.zip
Photo of Justin Gehring

Justin Gehring

  • 9 Posts
  • 1 Reply Like
My video card is an Nvidia-based 260 Card, latest drivers.
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
Ach, I hoped it was an AMD one, pointing in the direction of non well optimized support for AMD and relying on future upgrade. So ... TechSmith turn to move!
Photo of Shane Lovellette

Shane Lovellette, Employee

  • 257 Posts
  • 40 Reply Likes
I want to thank everyone for your thorough testing and feedback on this thread. You are fantastic! We will continue to look into this and hopefully figure out a more concrete answer. Your testing is very helpful!

Best Regards,

Shane Lovellette
Photo of Kevin Liu

Kevin Liu, Staff Software Engineer

  • 113 Posts
  • 21 Reply Likes
Hi Justin and Fabrizio,

Thanks for the test posted, I really appreciate it! To better understand the issue you have and speed up the communication process, I am going to contact you through email if that's OK for you.

Thanks,

Kevin
Photo of Fabrizio Camuso

Fabrizio Camuso

  • 73 Posts
  • 7 Reply Likes
It's ok for me, thank you.