Do not worry.

Nothing wrong with G1X.

And with about 99% probability, actual skipping and/or resizing happens on the sensor side.

While you are not impressed most people who'll buy this camera will be quite impressed. As it is quite high quality camera.

Canon have run out of technical excuses for crap video modes, maybe they still have the most important excuse of all - the consumer. Joe. Pa. Ma. Bob. And other idiots who will by consumer cameras without realising what makes a good video image.

IIRC Canon admitted the 5DII (ie: The Digic IV processor) could actually handle 1080p 60 fps, there is enough head room on the processing to implement even a basic bilinear or bicubic scaling filter.

Though they may end up with a bigger latency per line read (even more jello) if they can't grab the entire frame fast enough.

I think I remember they also saying they looked into faster line reading for less jello by using higher voltage, but it reduced battery life.

So I don't think it's anything to do with the Digic V, that'd definately faster enough, it's just no written in the firmware to do that. And I do think it's software, not hardware, I think everything points to that, it'd greatly increase hardware complexity and expense otherwise, and be kind of pointless.

Obviously they can both read full frame (stills) and choose where to read from (perhaps leaving power off to some locations they don't want).

Anyway if you read through this interview recently posted on here - http://www.imaging-resource.com/news/2012/01/15/panasonic-learning-from-the-fringe-taking-small-steps

You'd see the logic in not making it perfect for what the enthusiasts wan't, but standard/useable for everyone.

Again.

Obviously they can both read full frame (stills) and choose where to read from (perhaps leaving power off to some locations they don't want).

I didn't say the image processor reads the full frame then tosses out lines. Though I'd like to see where that 99% coming from still. This also doesn't preclude the fact, that almost certainly is still software controlled by code, and not by hardware, to select what/where to read from. Otherwise it is unnecessary expense, and complexity for more things to go wrong. I'm fairly sure you'll find there is only 1 circuit coming off the image sensor.

Plus if you had a 2nd circuit, it would be pointless, as you're still using the same ADC and the same image procesor (Digic IV), so I think there is no point to do that, as there is no benefit and no increased speed for the increased cost and complexity of a 2nd analogue circuit. As well as logically you'd think it'd be wired to read only the lines it needs for video, but it should be connected to the entire sensor in it's off state so that it doesn't have an impact on stills mode, which you would see the lines down the picture it would be connected to differ slightly to the lines it is not connected to. As well as noise there would be different again.

This is why I am saying it takes the same path from read out to image processor.

It also doesn't preclude the fact the full frame reading capability, and the ability to down scale with a filter like bicubic if it were implemented in the code.

Digic IV seems like it would be fast enough from what Canon have said to do it to process that much data down, but increased latency on the read out may increase jello, and there is the question of bandwidth between the ADC and image processor and how much the ADC can process before that bus as well.

@Vitaliy you are right to say sensor has a big role in downscaling but I think Athiril has some good points too actually.

We know how these things work now... The tech is well known...

The sensor in the GH2 has the A/D converters on board, so it sends a 100% digital signal to the LSI. This allows the first stage of the image processing pipeline to run faster. The faster it runs the more data gets off the sensor and into the encoder. Cinematic video is 24p and it doesn't wait for a slow sensor. A still can wait for a slow sensor and just take longer to write to the card. So video is more speed critical, and data throughput critical.

Therefore the DSLRs with the fastest sensors and fastest encoder chips produce the best image... No surprise there. The G1X is probably sensor-speed limited because it uses a similar sensor to the 7D. It isn't a next generation sensor, just a next generation LSI.

To get to 60fps in 1080p on the NEX 5N Sony must have taken all kinds of liberties with reading the sensor for video, which explains why that too is not up to the GH2 level.

So although Panasonic are not ahead on the stills quality of their sensor, they are ahead on speed, and that is why they win at video.

It isn't speculation, it is fact. You don't even disagree!

Seems good enough for what it is. Seems a pretty great camera to me (if overpriced) Why is that when a new photo camera comes along its being looked as a video tool? Not every camera has to have a brilliant video mode. Its a photo camera first and foremost. Does the fuji x100 have a great videomode? Does it really matter when there are other better tools solely for videography?

Just because the GH2 is a flash in the pan for video does not mean all others must be judged. If comparisons should me made then try seeing them as photo cameras first with video mode second and we all know how badly the GH2 fares in that area.

7d,5n,g1 x,g3,d5100 so on...They all shoot pretty capable,decent video...christ have we forgot how bad it was a few years ago?

Im sure there will be some ludicrous G1 x vs GH2 video comparison article on its way in the future.