QT screen recording - quality settings, huge files, strange codecs

Can anyone tell me if this is normal behavior?

Yes, it is normal behavior.

First of all, you are comparing apples and oranges here. Photo-JPEG is a medium efficiency, medium compression format whose data rate is determined by the quality setting and encode dimensions used while H.264 is a highly scalable compression format which employs a context adaptive approach to set encoding Profile and Level combinations which, in turn, limit the encoding dimensions and data rate allowed for a particular frame rate.

Quality: HIGH

Duration: 30 seconds

File size: 2.5 MB

Bit rate: 448

Codec: H.264

Quality: MEDIUM

Duration: 36 seconds

File size: 343 MB - yes really! - that would be 600 MB/sec

Bit rate: 75,031

Codec: Photo-Jpeg

Not only is this off the scale in terms of file size but - these are the DEFAULT and ONLY options provided! I have to assume this is a bug since there's no way Apple expects people to be making like 50 GB screen recordings. It's almost like the medium quality setting is tied to single frame screen images rather than an actual video file or something? A bit rate of 75,000 seems pretty off the charts too.

Your calculations are in error. 75,031 Kbps (75.031 Mbps) is actually about 9.38 MB/sec and when you multiply this by the stated duration of your file (36 sec), it comes to about 338 MB which is pretty close to the stated file size of 343 MB. As to "Default" settings, QT X encoding is predicated upon the use of presets which do not allow users to modify the encoding process. If you want to manually encode the screen, purchase third-party screen recording software which lets you record to an intemediate codec which you can then transcode manually to a codec/settings of your own choice or use an application like QT 7 Pro, MPEG Streamclip, ot HandBrake to custom re-compress your current files.

Is there a way to produce smaller files for screen recrodings - even the much smaller high quality files are very large when recording for 30 mins.

Files can frequently be made smaller but this often reduces the quality and/or resolution of the file. (I.e., the file size is directly proportional to the total average data rate, so any settings changes that reduce the data rate will make the file smaller.)

Lastly is there a way to reduce these files - perhaps soem transcoding to smaller versions - like on the order of less than 15 MB? I can export in QT player (used the iPad size setting) but that only cuts the size down by about half. I suppose I coudl try the iPhone size setting but that might not be legible to show what I'm doing on screen in the movie.

As previously noted, such files can be re-compressed. If you don't have QT 7 Pro, then you could use the free MPEG Streamclip app which offers similar transcoding capabilities. However, saying you want to create a 15 MB target file may or may not be realistic. For instance, from yout own posting, a 30 sec file would be easy to keep under 15 MB as long as you keep the total average data rate under 4 Mbps. But a 30 min file's total average data rate would have to be cut to 66.67 Kbps which, in turn would likely reduce your resolution to about 176x144 pixels @ 15 fps or 352x288 pixels @ 7.5 fps which, for me would not be a realistic target. Basically, sooner or later you will have to face the fact that if you want lengthy, good quality videos, they require large amounts of file space.

Perhaps what your technical description implies is that medium setting would be useful in certain instances - like perhaps if the screen recording area was a much smaller selection - rather than the whole screen? Or if I was using different display settings - lower rez, bit depth etc? Perhaps if one was recording with the intention to subsequently extract frames from the movie - like for screen shots for a print PDF where the jpeg frames would reproduce much better than the variable compression frames in the H.264 codec? I wonder why a user would select medium for screen recording.

Perhaps it's just that for screen recording the medium setting isn't very useful - but it's totally appropriate for other types of video recording. If that's true perhaps Apple should disable it for screen caps or at least not make it the default... It mostly just seemed odd that for such a user friendly, and options-thin app I was surprised this was the default behavior and seemed so counterintuitive. Usually lower quality means less data, while the reverse is possible it's not usually desirable.

No, what I am saying is that the Photo-JPEG codec preset is optimized for "medium" quality output and that the H.264 code preset is optimized for "high" quality output and that each codec has its own independent range of operating characteristics which makes it impossible to make the kind of simple comparison you wish to make here. For instance the Photo-JPEG codec is commonly used in a data rate range of 8 to 35 Mbps for SD through HD content. On the other hand, H.264 is commonly used in a data rate range of 64 Kbps to 300 Mbps. Certainly Apple could have created a High Profile Level 5.2 preset that could output video with a data rate as great as 300 Mbps that will match your idea of how the two codecs should work with regard to output file size—but why bother if "high" quality can be achieved at a much lower data rate setting by the H.264 codec which is many times more efficient than the Photo-JPEG codec. Besides, you are overlooking the fact that the choice of presets here is supposed to be based solely on the output quality desired—not the size of the file or how fast the codec encodes the data.

When comparing codecs, you must compare their characteristics in terms of three dimensional facits—time, space, and complexity. The Photo-JPEG codec is fast (in terms of time), moderate (in terms of space), and simple (in terms of complexity) while the H.264 codec is slow (in terms of time), compact (in terms of space), and complex (in terms of complexity). To do the sort of comparison you wish to make and get the results you expect, you must encode the same source file using the same codec at the same dimensions and frame rate but using settings optimized for "low", "medium", and/or "high" quality output. That is why I said you were comparing apples and oranges here. You are not comparing the "quality" aspect of the files created but rather their output size. There is a difference between the two.

Example:

Here is an H.264 "high" quality file which I will use ase the source file for conversion to the Animation codec. The "Anamation" codec is very simple and extrememly fast but creates huge target files.

Here is the Animation "high" quality version of the same file. You will note that this file is only about 1/60th as efficient as the H.264 file in terms of space for the same approximate level of visual quality.

And here is the Animation "medium" quality version of the same source file. Like the "medium" quality Photo-JPEG file in your query, this file is also much larger than my H.264 "high" quality source file—i.e., still only about 1/34th as efficient in terms of space but substantially smaller than that produced by the Animation codec when set for "high" quality. Thus, when comparing the "high" and "medium" quality file versions made using the same codec, the results are exactly as you expected—the "high" quality file is indeed larger than the "medium" quality file.

It would be great to have more control of the individual parameters in QT, rather than just 2 combinations of variables in the QT Player options but then I guess there must be reasons to upgrade to QT Pro or other apps.

Apple's basic approach is to simplfy recording/transcoding work flows to prevent users from making mistakes or creating video files which do not adhere to published playback standards. I personally do not rely on the built-in screen recording feature since its perfermace varies greatly depending on the content being recorded, the dimensions of the capture area, and the CPU power of the platform being used. Instead, I prefer a third-party dedicated app like Snapz Pro X which allows me to encode using a targeted codec (e.g., AIC, Animation, MPEG-4, H.264, etc.) depending on whether or not further processing is required for a particular project. (Screen Flow is another useful app if you prefer to capture and edit within a single app has an greater "learning curve" if you plan to take advantage of all of its built-in editing features.)

I still don't understand why they chose the set of parameters they implemented in the medium quality setting for screen recording but there must be use cases where it makes sense (like perhaps situations where less complex compression is required for faster decoding or for devices with slower processing capacity).

The Photo/Motion-JPEG compression format has a number of advantages and has long been a codec of choice for many older digital cameras. As you previously noted, each frame is the equivalent of an Intra-frame making easily editable at the frame level and quick to thumbnail for editing in video editors or GarageBand but not as compact as H.264 which employs both Intra- and "difference" frames which have to be referenced and expanded for thumbnailing, editing, and/or transcoding.

In any case the high setting makes for a nice recording and I can cook it down as needed. The 720p export option reduces file size from 200 MB to about 115 MB with acceptable quality for me and is suitable for playout by audiences' environments.

Apps like QT 7 Pro, GarageBand, MPEG Streamclip, HandBrake, etc. allow you to manipulate/limit a multitude of user settings which giving you much more precise coltrol over the final total average data rate which is directly proportional to the final target file size and output quality.