25F - stairs in thin objects

[Brandon Freeman]

LOL!! I love that!

[Philip Williams]

Originally Posted by Dennis Murphy

I had some stairs in my footage too.

Again, easily solved with a subtle blurring technique. If you scrutinize the frame, you may notice some softening of the image.

[Dennis Murphy]

Originally Posted by Philip Williams

Again, easily solved with a subtle blurring technique. If you scrutinize the frame, you may notice some softening of the image.

It's amazing what you people are doing with the new 33bit extrapolated algorerhythms.

[Bogdan Tyburczy]

Or just have a strong drink or a beer too many. Results may vary, but what you will see may also be similar to 33 bit... whatever :)

[Bogdan Tyburczy]

... and just want to add that 33 bit algorithm is perfectly reversible with proper sharpening, as seen in attached example ;-)

[Poppe Johansson]

Originally Posted by Dennis Murphy

I had some stairs in my footage too.

Uh.. really bad looking stair-stepping you have there. Haven't ever seen that bad in my shots :D

[Alvise Tedesco]

Completely off topic and random questions, but I like the seriousness you approach your camera:

1) How do you protect for overexposure, since zebra didn't show 100 ire. Just guess once used to how lcd behave?

2) How do you like your Cineform+FCP workflow? (..well, maybe nobody use it since it would probably involved in stair stepping issues conversation)

For whom had previously a DVX:

2) How do you like/dislike focusing ring feel?

3) Better sd (also in 4/3) now or before (intentionally not talking about 16/9 where advantage has to be obvious)?

Cheers

[Tony Tremble]

The stair steps are cause by the way the Canons create progressive frames from interlaced ccds.

The XL-H1 and XH camera clock the greed CCD a field out of sync with the blue and red CCDs like so.

rbrbrbrbrbrb
gggggggggg
rbrbrbrbrbrb
gggggggggg

This obviously means there is a green sample or a lower luminance same in the field sampled by the red and blue CCDs hence the slight loss in resolution seen as stair stepping round areas of high contrast.

TT

[Mikko Lopponen]

Originally Posted by Tony Tremble

This obviously means there is a green sample or a lower luminance same in the field sampled by the red and blue CCDs hence the slight loss in resolution seen as stair stepping round areas of high contrast.

So it creates the same artifact as your basic hdv-deinterlacing would? Hmm. Not good.

[Bogdan Tyburczy]

Originally Posted by Mikko Lopponen

So it creates the same artifact as your basic hdv-deinterlacing would? Hmm. Not good.

I think Canon's technology is much better than that. Images speak for themselves. XL-H1 and A1/G1 produce the best images below $10k imo and do not exhibit painful problems that software deinterlacers do.

[Tony Tremble]

Originally Posted by Mikko Lopponen

So it creates the same artifact as your basic hdv-deinterlacing would? Hmm. Not good.

That's not what I was saying at all. The technique Canon use is no different than how green ccd shift processing works that you should already be familiar with. Think of the system as temporal green ccd shifting.

The Digic processor can easily construct information from adjacent ccd samples. The limiting factor with this method is when a high contrast area is sampled and there is an absence of red or blue information thus there is a resolution drop which appears as stair stepping.

It is actually very good.

25F has much more resolution than a deinterlaced 50i stream. Other cameras in the XH-A1 price range suffer from more fundamental problems of either not having much native resoluton or have poorly implemented HDV codecs.

Message to Chris Hurd.

I can hardly access this site any more. I keep timing out and get sent to the archives section. Can you help please???

TT

[Mikko Lopponen]

Originally Posted by Tony Tremble

That's not what I was saying at all. The technique Canon use is no different than how green ccd shift processing works that you should already be familiar with. Think of the system as temporal green ccd shifting.

Yes, I know the theory but shouldn't that just result in lower res images not stairstepping? I mean the stairstepping in those example images is exactly the size of one line. The exact same size of stairstepping is created by deinterlacing hdv footage. High contrast images also show regular deinterlacing way better so it doesn't necessarily have anything to do with not enough colour.

25F has much more resolution than a deinterlaced 50i stream.

How much? Is it possible to get a 50i screengrab vs the same image in 25f?

[Paul Firth]

I have to agree with a previous post--I think all the stair-stepping is due to the sharpening that the camera does before saving to tape. If you look closely, the black stick is blacker and the sky is whiter where they meet each other--this is a clear symptom of software adding contrast (even though it is done in-camera). Try turning your sharpening down to -9. I have done that with my A1 and the pictures retain all of the sharpness of the original image without the "stair-stepping." Then you can add a little sharpening in post if desired.

[Brandon Freeman]

The thing with that though, is that the stair stepping will reappear if you apply sharpening in post, and much worse. I think the sharpening is honestly 0 at 0, and -9 simply adds a blur. Honestly, Canon does such a great job, and I have not seen stair stepping to be really an issue at all. I have noticed it in some shots, but it doesn't merit concern or panic. At least for me. Again, when downrezzing to 720p, it's gone.

[Tony Tremble]

Mikko

I have got some grabs here that compare 25F to 50i to 50i_de-interlaced and categorically show how effective 25F is.

You'll have to take my word for that 25F has significantly more resolution than a de-interlaced 50i image. In order of resolution 50i has the most followed closely by 25F then a long way behind is 50i_DI.

Unfortunately I am not allowed to post attachments. Looks like I am still sat on the naughty step.

All the best

TT