Recently, I came to learn from Harm Millard of ppbm7.com that Adobe Media Encoder was considerably slower than exporting directly from Adobe Premiere Pro or Adobe After Effects. Since I currently have both CS6 and CC (Creative Cloud) installed, I thought I’d give them a spin and test it myself.
Important: This isn’t a scientific test, but just a quick one I use practically. It’s similar to how I test workflows prior to rendering projects, and has served me well over the last five years. You must test your own workflows on systems you intend to use it on, to know if these results are applicable to you. Don’t assume my results and findings will be similar to your own.
My testing setup
I used my office computer, a 27″ iMac i5 3.2 GHz quad-core with 8 GB 1600 MHz DDR3 RAM and an Nvidia GTX 675MX 1 GB GPU. The OS was OS X 10.8.4. This computer is not an editing machine, but the results should be typical for most single-CPU-multi-core systems.
On all softwares, Mercury Playback Engine was turned on, in hardware mode. Compression, though, relies heavily on CPU and RAM.
I’m pretty sure many Internet and background processes were running while the tests were conducted. I had both CC and CS6 versions of Adobe Media Encoder, Premiere Pro and After Effects open at the same time. I tried as best as I could to ensure no undue processes were running that would skew the results, however, on a ‘cleaner’ and more optimized system, the results should be better and more consistent.
The two tests, and the goal
The goal of the test was to determine if there is any sizable difference in rendering times between:
- Direct exports from Premiere Pro or After Effects, in both CC and CS6.
- Exporting via Adobe Media Encoder from After Effects or Premiere Pro, in both CC and CS6.
In plain-speak, I wanted to know if CC is better than CS6, and if AME is better than exporting directly.
These tests are what I’ve been running ever since I got into compressing my own videos. The first test is a simple export of a clip from one format to the intended delivery format. In this case, the specifications are as follows:
- Original Clip – 11.01 seconds (276 frames), 1920x1080p25, 8-bit 4:2:2 in the Canon XF codec (MPEG2 422 Long GOP), MXF Op1a wrapper, Canon C300 native footage at 50 Mbps. Audio was 16-bit 48 KHz, and the total file size was 68.2 MB.
- Output format – 1080p25 8-bit 4:2:0 H.264 Quicktime. Audio is AAC at 48 KHz. Total bit rate is limited to 10 Mbps.
The second test is more indicative of real-world results, and is as follows:
- Mixed clips on timeline, both Canon XF and H.264 (DSLR) material, edited on multiple tracks. All other settings same. Total length: 820 frames.
- Output format – Similar to above.
Additional codec settings are as follows (Maximum render quality/bit depth rendering was not selected):
In the case of the first test, the exact same clip was exported from all the softwares. In the case of the second test, the sequence was created in Premiere Pro and exported. The same sequence was imported in After Effects and exported.
Each test was conducted three times, and the results you see below are the average scores.
I have avoided filters and effects, or anything that might interfere with the rendering times that do not pertain to actual compression. Effects in different versions of Premiere Pro or After Effects might not be same algorithm, and it wouldn’t be fair to judge them directly under this simple test.
Here’s a table of the results of test (all values in seconds, lower is better):
|Direct||Adobe Media Encoder||Difference|
|Premiere Pro CS6||Test 1||16||16||0|
|Premiere Pro CC||Test 1||16||14||2|
|Difference between CS6 and CC||0||2|
|After Effects CS6||Test 1||38||24||14|
|After Effects CC||Test 1||28||28||0|
|Difference between CS6 and CC||10||-4|
Observations and Conclusion
There is hardly any difference between Creative Cloud or CS6, with or without AME.
On the other hand, within each software, it is much faster to export via Adobe Media Encoder, because the difference is pronounced.
Conclusion? Doesn’t matter which version of Premiere Pro you have, but no matter what, try to export via AME and not directly.
Winner: Adobe Premiere Pro CC with AME!
Things are more interesting here. The big red scores at the bottom clearly show that Adobe After Effects CS6 is faster at exporting than After Effects CC. There is one big exception. In the simple first test, CC performs better than CS6.
As far as AME is concerned, in both CS6 and CC, AME performed far worse than exporting directly. I must say, based on this, that I would not export a project from After Effects via Adobe Media Encoder. Add to render queue and export directly for the fastest results.
Harm’s findings are as follows:
…using Direct Export is almost five times faster than using the AME Queue…
My findings are different. As far as After Effects is concerned, the difference between direct export and AME is roughly 35% (1.35x). It is not a negligible difference.
Winner: Adobe After Effects CS6 direct export!
What do you think? Are your results similar, or are they totally different?