Im totally new to video camera stuff. I just bought a FX1 and was wondering how do I get it to record in Cineframe 24 mode. I can't seem to find it.

The CineFrame modes can be found in the "picture profile" settings. You have to first hit the "picture profile" button, then select a profile to edit, then scroll through the settings until you find the Cineframe modes (which are near the bottom).

CF24 can also be accessed by the default PP4.

-Steve

Don't do it man! Save yourself, look on some of the other posts. Cineframe degrades the image to a great degree, IMHO it shouldn't even be allowed on the camera. The highly compressed editing process is going to down the quality of the original image quite enough so you don't want to start on a bad foot. If you want the 24p look (and at at least half of us here do desperately), do it in post.

I disagree Betsy.

The Cineframe modes offer an improvement in colour space for progressive images, and an ease the compressor.

Yes, there is a sacrafice of resolution that the CF modes entail, and yes, there are some issues with CF24 motion cadence (though not CF25 or CF30), but there are gains in quality and in workflow that should not be discounted.

A lot has been written about the CF modes that is misinformed or poorly contemplated.

-Steve

Well, Steve for one thing the proof is in the pudding and every finished product I've seen using cineframe is uh... well... bad pudding--every time I see it I feel like I need glasses or I'm tired or I have a migraine--it's just slightly... blurry. So frankly I don't think the anti-cineform stuff has borne out to be misinformed at all. The resolution loss is significant and why did we wait so long and pay so much for these cameras if we're going throw their one killer app away? Add to the fact that some (many?) people are intrigued by cf24 because they have dreams of some day making a masterpiece that'll someone will buy and transfer to film and they want to record to cf24 because it has that "film stutter look" and then they find that film transfers don't do very well at all with those programs. And yes, compared to cf24, cf25 and cf30 are slightly better but that's like saying for a fat girl you don't sweat too much. Unless you have some very specific purpose in mind I'd avoid all those modes like the plague and be vary wary of advising some other poor soul to walk down the path to mediocrity.

That's some fancy emotionally evocative language you have there.

Let's just say HDV as a format is on the low end of HD production - so you're already on the way to mediocrity, and your objective is to get good progressive imagery at a 25 Hz or a 30 Hz frame rate. (I'll ignore 24 Hz, since nobody will believe anything on that subject - you'll have to see it for yourself). I'll do the rest of the argument at CF25 to "push" the "filmic" aspect.

Now, your storage medium is 4:2:0 1440x1080 MPEG-2. You have a choice: you can either shoot 50i or CF25.

At 50i, every field you acquire collects an the opposite Cr or Cb channel. When you go to (adaptively or otherwise) deinterlace, the Cb channel of one field will be derived from the Cb channel of the opposite field. Vice versa for the Cr. In situations of high motion, not only will your compressor have to compress each field with significantly different luma infromation (requiring more bandwidth), but the half the chroma of each field will be offset by the temporal difference between frames. This temporal offset will make keying and other post production difficult. While an adaptive deinterlace will result in superior resolution in rather still imagery - if you were shooting stills, maybe you should have bought a digital SLR.

Suppose you deinterlace before compressing (i.e. CF25). The actual resolution of the sample is only 1440x540 (notice this is more than 2x the resolution of SD, and is 85% of the resolution of 1280x720p and still 112.5% the resolution than DVCPRO-HD 720p offers). You also have the whole 25 Mbps to encode this 1440x540 worth of information. In other words, the CF25 content is 50% less compressed than the interlaced signal. When you upsample and plunk this into 1440x1080 4:2:0, the Cr and Cb channels come from the same temporal sample regardless as to whether the actual MPEG-2 encoding is flagged as interlaced (which it is) or progressive. This will be advantageous to keying. Notice also, because there is less real information in the image, the image is softer (migrane inducing to you), however - the reduced bandwidth of the image also reduces strain on the compressor (anthropormorphically giving it less of a migrane) - and this lower resolution image is therefore recorded more accurately. The result is that as opposed to having a 1440x1080i 4:2:0 image, you really have a 1440x540p 4:2:2 image stored in a 1440x1080i 4:2:0 container.

Undoubtedly if I was looking for a "reality" feel, I'd go 60i/50i and enjoy the added resolution... but if I was looking at web-release, progressive DVD, or a lot of post production and compositing... even a 720p "master" - I wouldn't think twice about using the CF formats.

Also, if you can walk up to anyone still shooting 720x576 4:2:0 (or 720x480 4:1:1) miniDV and tell them you have a 1440x540p 4:2:2 camera - do you think they'd question it as an improvement? (rhetorical - I know there are significantly more detailed answers for that, since I've read all the skepticism) That downsamples very well to 4:4:4 SD you know.

I'm sorry if you don't like any of the CF25 material you've seen... but where have you been looking? I've seen some CF25 material I've very much enjoyed.

-Steve

Thanks for the info. I woke up this morning and looked in the book a lil harder and found how to get the cf24 mode working. Figured they would have it in big bold print. I just want to see what it looks like playing and after I capture it. Might create a unique look after some post work.

I got a lot of questions but I'll see if i can find them myself 1st.

Thanks.

Betsy you should take a look at this Cineframe25 footage here:

http://personales.ya.com/autodrome/

After seeing Cineframe24 I thought all Cineframe was garbage, this footage promptly put me in my place.

"At 50i, every field you acquire collects an the opposite Cr or Cb channel."

4:2:0 in interlace does not work like that. Cb and Cr are on alternate lines of the field, not on alternating fields.

4:2:0 in interlace is bad, but not quite as bad as you're making out :-)

1080i60 from the FX1 / Z1 downconverts great to 720p24, or 720p30.

Graeme

4:2:0 in interlace does not work like that. Cb and Cr are on alternate lines of the field, not on alternating fields.

Thanks for correcting me! I can't emphasize how much my interpreation of 4:2:0 has bothered me. So for interlaced is it:

Cr
Cb
Cb
Cr
Cr
Cb
Cb
etc?

On my stress-tests of HDV I knew somehting was wrong with my intepretation beause the colour fidelity of each field wasn't half bad.

-Steve

Something like that, it's pretty hard to describe :-)

http://www.poynton.com/PDFs/Chroma_subsampling_notation.pdf

Might help

Graeme

this is sorta off-topic but... Tony, THANKS FOR USING COLDPLAY MUSIC in your preview video! I love Coldplay!

You have a choice: you can either shoot 50i or CF25.

I'm sorry if you don't like any of the CF25 material you've seen... but where have you been looking? I've seen some CF25 material I've very much enjoyed.

-Steve

Both, as you say, involve compromises. CF25 is very nice for making PAL DVDs where the loss of vertical resolution won't noticed -- or for the web.

On the other hand, film labs have been converting 1080i50 (and 1080i60) to film for years and so the process is well established. Plus, you get a full rez HD video master.

Frankly, I would never do anything to compromise my Master tape based upon someone's claim that "4:2:0 interlaced won't look good when deinterlaced."

IF that's a real problem, there are a lot of smart software engineers who will attack the problem and likely come-up with a fix.

One such person just responded to your comments! :)

What's the story Graeme?

4:2:0 deinterlaces fine if you're very careful. I'd not worry as long as the software is competently written.

I favour, with the Z1, a 50i shoot, with a 25p deinterlace, then conform to 23.98fps for "best" results, and best flexibility (50i makes much better slowmo than 25p) in post.

Graeme

However, it is more the way they have handled the focal length what makes it look more film like, in addition to the lighting and frame rate.

That Guerilla35 adapter with Nikon lenses is what makes it more filmlike.

Anybody in this forum has ever used that adapter???

It really affects the focal length dramaticaly. Not even using the long telephoto setting with large apertures can provide the look with the FX1 or Z1 . You really need an adapter of that sort with 1/3 inch cameras.

Betsy you should take a look at this Cineframe25 footage here:

http://personales.ya.com/autodrome/

After seeing Cineframe24 I thought all Cineframe was garbage, this footage promptly put me in my place.

this is sorta off-topic but... Tony, THANKS FOR USING COLDPLAY MUSIC in your preview video! I love Coldplay!

Uhhh..wasn't my clip, just merely pointing to some great footage shot using Cineframe25. Hope my previous post did not imply that the clip was mine.

I do, however, enjoy Coldplay.mine.

Hello. I just read this post. And i agree with what Steve says. Maybe HDV is not the best HD codec for going up to 1080 but i tell you that recording at CF25 you get a BIG difference with "old" miniDV when going to PAL.
Color correction is infinite and the extra detail quality really enhances the images after you treat them.

I am just finishing a project that was done entirelly at CF25 because i thought the compression ratio is much better and also it will be easier for the editing (less process time, no one says this but it is important when we do not have real time video cards). Also when i want to do a slow motion i just switch from P1 to P2 settings. They are the same except for the CF, so i can quickly move from one to another.

This links will be online in an hour:

http://personales.ya.com/autodrome/mardentro.mov
http://personales.ya.com/autodrome/escaleras.mov (in this clip one shot was done with dvx100e, guess which one)

Graeme said about MPEG 2 4:2:0 Interlaced: "it's pretty hard to describe :-)"
(a big thank you to him and also to Steven White for his explanations)

I nevertheless would like to dig to the bottom of it. I have not found one conclusive source on the web, but this forum is full of brilliant experts.

There are based on my understanding two possibilities for colour subsampling to work 4:2:0 interlaced: (each element = 1 pixel)

Version 1)

YCb Y YCb Y YCb Y ... (1st line Field 1)
YCr Y YCr Y YCr Y ... (1st line Field 2)
Y Y Y Y Y Y ... (2nd line Field 1)
Y Y Y Y Y Y ... (2nd line Field 2)
YCr Y YCr Y YCr Y ... (3rd line Field 1)
YCb Y YCb Y YCb Y ... (3rd line Field 2)
Y Y Y Y..... and so on

alternatively:

Version 2)

YCb Y YCb Y YCb Y ... (1st line Field 1)
YCb Y YCb Y YCb Y ... (1st line Field 2)
Y Y Y Y Y Y ... (2nd line Field 1)
Y Y Y Y Y Y ... (2nd line Field 2)
YCr Y YCr Y YCr Y ... (3rd line Field 1)
YCr Y YCr Y YCr Y ... (3rd line Field 2)
Y Y Y Y..... and so on

This is inferred on my knowledge of 4:2:0 progressive and some logic reasoning and video experience. Which one is correct? Version 1 or 2?

If 1 is correct, cineframe 30 or 25 lead to half the resolution (540 lines) but with 4:2:2 colour spacing.
If 2 is correct, cineframe is indeed fairly useless, as it adds very little extra compared to dropping a field in post.

This is very important to me as I am thinking of buying/renting a FX1 for chroma key work (and I am in Pal land and would be ok with 1440*540)

Your help is very much appreciated - and the results probably very interesting for other FX1 users as well