Sean Oneil
Forum Replies Created
-
[JeremyG] “Sorry Sean, that method is totally not what is going on with the Kona.”
Do you have a Kona 3?
-
[JeremyG] 60p frames converted to 60i fields (which is 30 frames per second)
With FCP conversion yes. It becomes 30 frames per second. With the Kona conversion, no. It remains at 60 frames per second, but half of the resolution is gone because one of the “would be fields” for each frame is thrown out.
Think of it like this. Start with 720p60. Pretend the first thing that happens in both methods is that it is simply uprezzes it to 1080p60. So now there is a 1080p60 image. From there, the Final Cut method will take that and convert to 1080p30 by throwing out every other p frame, and then interlace the thing making it 1080i60 – 30 frames per second.
The Kona would take that 1080p60, and interlace it as-is making it 1080i120 (yes, 120 fields). Then it will throw away every other field, bringing it to 1080i60 – 60 frames per second.
[JeremyG] “True 720p60 material is 60 discrete frames per second, not 30 doubled frames (unless you shot with a varicam @ 720p30 over 60).”
Obviously. Isn’t that the foundation of the entire discussion? -
[JeremyG] “It’s two fields of 540 lines that are 1/60th of a second apart that when combined make up a 1080i frame.”
Well after the conversion there’s only one field so there’s nothing to combine. The Kona conversion omits the other field. So you are left with sixty “half-frames”, which when played back properly becomes 540p at sixty frames per second. The full 60fps is maintained but it is only half the resolution of 1080 lines (although in the case of converting from 720p, it would be more like 2/3rds the resolution). Note that the FCP Canvas uses the same method during playback – it omits every other field (cutting the res in half) so you never see jaggies. Does it with SD as well.
The FCP 720p-1080i method is completely different. Instead of cutting the resolution in half by chucking away one of the fields, FCP instead chucks away every other frame (like I thought everything did) – but keeps the full 1080 lines for each frame. So it is essentially 1080p at thirty frames per second. Higher res but half the framerate.
I see wisdom and folly in both methods. It’s just robbing Peter to pay Paul or vice-versa. I do wish the AJA Control Panel allowed both options, and I wish the FCP preferences did as well. And I wish the manuals for both products would explain this stuff so that you and I wouldn’t have to waste time discovering it ourselves. Anyway, I feel smarter now. Good to learn new things.
-
After all that, I stand corrected. To my shock the Kona 3 does convert it the way Jeremy described. I completely understand the situation now. Sorry for the attitude and pardon my ignorance. This is how I learn things. It has been a really nasty day and I’m on some kind of weird medication (seriously). But I still stand by my assertion that the FCP6 method is superior.
The Kona 3’s method preserves the temporal quality at the cost of spatial quality. It keeps it at 60fps by literally chopping the resolution in half. So while you maintaining the 60fps, you only get 540 lines of video instead of 1080. The idea is that it gets played back at 540p60 on a consumer television after being processed by the video decoder. I think it would almost be more technically appropriate to call it 540pSF instead of 1080i. This video cannot be presented in interlaced format. It must be processed/deinterlaced. If you watch it on your broadcast monitor you’ll notice how awful it looks without being processed.
FCP, Motion, and AE instead preserve the spatial quality. So you get a full 1080 lines of video per frame. But it must cut the frame rate in half and make it 30.
In other words, this is a creative decision. A matter of taste or preference. I’d bet a Teranex allows both options. But it is my strong opinion that the FCP method is better and safer, and I think it’s pretty lame that the Kona can’t do either or. I’m not sure all consumer equipment can properly decode this 540psf-type signal that the Kona creates. Maybe it does. It should be noted that the FCP method is much more difficult to process, since it involves upscaling to 1080p first, then throwing away ever other frame.
-
I just did it w/ a Kona 3. I don’t have an HD deck in house, so I just fed the SDI to another NLE. 720p timeline, used the Kona control panel to do a 1080i cross conversion.
On the other machine, I just captured 1080i using capture now. I’ll post the QT in just a few minutes.
-
[walter biscardi] “What you’re seeing here is the result of FCP not properly converting 720p/60 to 1080i/29.97.”
Wrong. Why don’t you guys post clips of the Kona 3 doing the conversion. Then we can all see how it does it the same exact way.
-
[JeremyG] “Okay, that’s FCPs way which I wouldn’t trust as working as any other converter in the world. FCP lays out 24p material on a 29.97 timeline as 2224. “
Don’t want to toot my own horn, but I am the one who originally pointed out the 2224 thing with version 6. I also pointed out how it scales between 480 and 486 instead of cropping (a bug which didn’t exist in FCP 5). FCP6 doesn’t convert everything right. But in this case it does.
I can PROMISE you this is the way 60fps 720p is meant to be converted to 1080i. No other way. Think about what the image would look like on a display if it worked the way you described. It would blend mismatched fields together. Pure filth. Might as well make a 3/4″ SD tape because it wouldn’t look any better. Trust me, it doesn’t work that way.
If you do find a converter that does it this way then it is wrong, and FCP6 is right. But I’m 100% certain you will find any converter that does it that way. Not a Kona, not a Teranex, nothing.
-
[walter biscardi] “You can. Open the post. Edit Post is in the lower left. You have 30 minutes from when you first enter the post or until someone replies to it.”
Thank you (and Jeremy) for that. I’ll be sure to use it!
-
Proof.
Here is an “original” 720p 60 frames per second video clip. Timecode generator showing you the frame count:
https://www.geocities.com/lifterus/Original_720p60.mov
I then exported that file as a self-contained quicktime, re-imported it, and placed it on a 1080i sequence. Rendered, output it, and this is what it looks like now:
https://www.geocities.com/lifterus/Converted_to_1080i.mov
That is a true 1080i60 interlaced quicktime video. Notice how every other frame has been omitted. That is how it works.
Had it worked the way you described, you would see timecode numbers blended together.
-
[JeremyG] “You have 720p60. Each frame of 60p gets put into a field, which means you turn 60p into 60i which is 30 frames per second. “
You’re contradicting yourself in the same sentence. It does not pull a field from each frame in this case. If it did you wouldn’t have 30 frames per second. You would have 60 frames per second, but half of each frame would be missing.
Maybe your source is not actually 60 frames per second. That would explain why you think you are seeing that. Most 720p60 content is really only 30 frames per second (each frame is doubled up). In this case there’s be no way of identifying it.
This is going nowhere. I’ll post some quicktimes in a bit to show you.