Out of devilment I went on to the Apple site, and tried to configure a new iMac for very high performance and storage. It's possible to get the price to over £3k for a Retina display 27 inch model with maxed out memory etc.
What I'd really like to know is about the performance of the storage nowadays. According to Apple, their latest SSD storage is considerably faster than their previous generation systems. However there is a size limit of 1 Tbyte. On the other hand, there is a 3 Tbyte Fusion drive option.
I have some experience of SSD drives over the last few years, but only limited experience of the Fusion drives. For really heavy duty work (complex video editing say) is SSD the only way to go, or would Fusion drives give almost the same performance at less cost, and provide more significantly more storage? Fusion drives may work well enough for every day tasks, but not for some really large ones. I don't know - not yet, anyway.
Maybe nobody here has tried the kind of video processing which would reqiure such computer power.
It may also be the case that software could have a very significant impact, with different software packages performing more or less well. One task which I tried to do recently was to remove the shakes from a 12 minute HD video. The actual processing on my fairly fast machine took about 12 minutes, but the results are not particularly good. I don't even know if it's possible to get good results, but it might just work out if the parameters are shosen well, or at least better. Conceptually it ought to be possible with a careful alignment algorithm over successive frames, but with 1800 frames per minute, the 12 minute video would have 21600 frames to check - each in HD. If it was vital, then brute force manual work at say 15 seconds for each frame alignment would take 450*12 =
For a task like that it might make sense to split the whole video into short sections and then try the processing on small sections to evaluate how well it all works. Then either all the sections could be processed separately then joined together, or if a good set of parameters is found, the whole video could be processed with that set applied.
It may also be the case that some software for reducing camera shake effects is much more effective than others. My own immediate needs are limited, and probably would not justify spending a huge amount of money to get a very high performance machine. I do have some videos I'd like to clean up, if possible, but there are other solutions - including just not bothering on cost and time grounds!
Another "solution" would be to wait a few years, and exploit Moore's "Law" - performance/price ratios could be 10 times better in 5 years time!
What I'd really like to know is about the performance of the storage nowadays. According to Apple, their latest SSD storage is considerably faster than their previous generation systems. However there is a size limit of 1 Tbyte. On the other hand, there is a 3 Tbyte Fusion drive option.
I have some experience of SSD drives over the last few years, but only limited experience of the Fusion drives. For really heavy duty work (complex video editing say) is SSD the only way to go, or would Fusion drives give almost the same performance at less cost, and provide more significantly more storage? Fusion drives may work well enough for every day tasks, but not for some really large ones. I don't know - not yet, anyway.
Maybe nobody here has tried the kind of video processing which would reqiure such computer power.
It may also be the case that software could have a very significant impact, with different software packages performing more or less well. One task which I tried to do recently was to remove the shakes from a 12 minute HD video. The actual processing on my fairly fast machine took about 12 minutes, but the results are not particularly good. I don't even know if it's possible to get good results, but it might just work out if the parameters are shosen well, or at least better. Conceptually it ought to be possible with a careful alignment algorithm over successive frames, but with 1800 frames per minute, the 12 minute video would have 21600 frames to check - each in HD. If it was vital, then brute force manual work at say 15 seconds for each frame alignment would take 450*12 =
For a task like that it might make sense to split the whole video into short sections and then try the processing on small sections to evaluate how well it all works. Then either all the sections could be processed separately then joined together, or if a good set of parameters is found, the whole video could be processed with that set applied.
It may also be the case that some software for reducing camera shake effects is much more effective than others. My own immediate needs are limited, and probably would not justify spending a huge amount of money to get a very high performance machine. I do have some videos I'd like to clean up, if possible, but there are other solutions - including just not bothering on cost and time grounds!
Another "solution" would be to wait a few years, and exploit Moore's "Law" - performance/price ratios could be 10 times better in 5 years time!
Comment