View Full Version : PXW-X70 - Moire from "line skipping" in 4K???


Cliff Totten
July 6th, 2015, 10:37 PM
Just got my X70 4k upgrade today and started test shooting.

I'm used to seeing the AX100 UHD/4k and have have shot easily 100+ hours with it. So I know the moire and aliasing it produces. (some high frequency "buzzing" but it's not too bad)

So today, I was playing with the X70 and it's industry rock-bottom 60Mbp/s UHD. I "think" I'm seeing more aliasing and moire in it than I do with the AX100. Again, this is just simple "eye balling" view and I have NOT done a true side by side test yet. Admittedly, I could be WAY off and completely wrong here but I'll just toss this up in the air to see what y'all think....

The Handycam AX100 is marketed as a "full pixel readout" with "no line skipping" in that camera's sensor readout process. However, this claim has not been stated anywhere about the X70's new 4K recording yet. Is it possible that Sony "line skips" on the X70 to reduce the data and processing power needed? (saving more CPU headroom for 4k encoding?) I'm guessing that Sony is struggling with the X70 4K recording for some reason. The option took a long time to arrive and only 60mbp/s was promised instead of the 100Mbp/s that every other camera uses. (still a huge mystery to many of us)

I know that I'm going way out on a limb here and I'm not accusing the X70 of line skipping at all. I completely acknowledge that it could very well have the exact same sensor read out as the AX100. I'm really just wondering out loud, that's all.

So, does anybody else notice more aliasing or moire from this X70 4k upgrade? (more than the AX100?)

I'll do better comparison later this week. I don't know yet, maybe it's the same. I just got a weird impression in my first 1 hour of playing with it.

CT

Noa Put
July 7th, 2015, 04:38 AM
You have both camera's so that would be very easy to find out, no?

Lou Bruno
July 7th, 2015, 06:14 AM
Let me jump in here with a non-techie observation. I own both the X70 and the AX-100. No update for me on the X70. However, I updated the free 100Mps firmware for the AX-100. I can see/observe less interline Twitter with the higher sampling rate vs. the 60Mps, especially horizontal and verticals.

I am sure SONY will increase the bit rate. Makes no sense to why a consumer camera would have better qualifications. .

David Heath
July 7th, 2015, 05:32 PM
So today, I was playing with the X70 and it's industry rock-bottom 60Mbp/s UHD. I "think" I'm seeing more aliasing and moire in it than I do with the AX100. Again, this is just simple "eye balling" view and I have NOT done a true side by side test yet. Admittedly, I could be WAY off and completely wrong here but I'll just toss this up in the air to see what y'all think....

The Handycam AX100 is marketed as a "full pixel readout" with "no line skipping" in that camera's sensor readout process. However, this claim has not been stated anywhere about the X70's new 4K recording yet. Is it possible that Sony "line skips" on the X70 to reduce ?

I think it line skipping is highly, highly unlikely for several reasons. First and foremost is that if the AX100 does a full readout, and the X70 does any form of line skipping, you are unlikely to be in any doubt about it. :-) Expect a pretty severe difference!

It's also worth thinking exactly what line skipping means, and what it implies. In simplest terms, it means not reading all the photosites on a chip to (as you say) "reduce the data and processing power needed". But you can't simply just skip any old number of lines - it has to be done on a regular geometric basis. There are other constraints as well - you couldn't (say) just skip every other line or you end up with green pixels and EITHER red or blue. Think of the Bayer pattern. Consequently one approach which does get used is "read two, skip two", and this has been used in combination with ding the same to columns, which maintains symmetry. So on average only a quarter of the pixels are used.

Now I believe for the X70, the effective 16:9 frame has about 14 million pixels available, so even the simplest form of line skipping is likely to reduce that to 7 million pixels, and only 3.5 million if it wants to maintain symmetry. Which is nowhere near enough for a decent 4K image. Do any such approach and you won't be left in any doubt that something not very nice is happening.

The only way to really get any true idea of what is happening is with proper test charts. Such instantly show up "secrets" of the internal tricks of how cameras do their work, and line skipping is a good example of that. If you want an example, then look at the charts shot with the FS700 at higher and higher frame rates. Move above the standard frame rates and it's obvious exactly what is happening.

Cliff Totten
July 7th, 2015, 07:43 PM
I completely agree with what you have said above. The line skipping is just a simple first thought that popped into my head on first inspection. I'll put both cameras up on charts this weekend and compare.

As far as RGGB Bayer, there are different ways to skip vertically and horizontally and of course, bin and sum values too. Very few people (only directly involved sensor engineers?) at Sony know that exact math. It's all in the "secret sauce" processing.

I just had a weird feeling when watching the first 20 min of 4k footage. I'm so used to seeing the AX100 that the X70 just struck me as looking as though it has more aliasing.

I'm not taking any of this seriously until I see the chart tests.

Hopefully I'll see the exact same results. ;-)

CT

Edit: There are tons and tons of Sony's very proud statements about the AX100 not doing pixel binning or line skipping. Sony is really touting and bragging about it. (and rightfully so) It's in all Sony marketing materials everywhere. However, oddly enough, nowhere in ANY Sony X70 materials is the same claim about the X70. How odd. Full pixel read out SHOULD be a huge selling point for the X70 too. Yet it's mysteriously missing.

Can anybody find ANYwhere that Sony states the X70 uses a full pixel readout with no line skipping or binning? Again,..I'm not saying the X70 doesn't process it's sensor readout the same way the AX100 does. But the fact that they don't to even mention that great feature in any way is kinda odd....no?

David Heath
July 8th, 2015, 12:39 PM
As far as RGGB Bayer, there are different ways to skip vertically and horizontally and of course, bin and sum values too. Very few people (only directly involved sensor engineers?) at Sony know that exact math. It's all in the "secret sauce" processing.
Well.... that's true to an extent, but quite a few things are fixed, so the options open to the designers are maybe far more limited than you'd think. And a relevant chart can go a long way to unravelling what secret sauce has been used.

The Bayer nature and photosite count are things that are fixed, as for that matter is the output resolution. Likewise the fact that a line can either be skipped or not skipped, and read lines contributing to each output pixel have to have all three colours. (Relevant as alternate lines will have either green and red, or green and blue photosites - hence the comment about not simply being able to skip every other line.)

It follows that the situation tends to be quite black and white when you're talking about 14 Megapixels - either it does a full deBayer and scale - or it lineskips. But if the latter, it can't "slightly lineskip", it's all or nothing, and pretty certain to be very obvious in a direct comparison, and even on ordinary pictures, let alone charts.
Edit: There are tons and tons of Sony's very proud statements about the AX100 not doing pixel binning or line skipping. Sony is really touting and bragging about it. (and rightfully so) It's in all Sony marketing materials everywhere. However, oddly enough, nowhere in ANY Sony X70 materials is the same claim about the X70. How odd.
I hear what you are saying, but there's a big gap in many organisations between the technical departments and sales/marketing. I'd be very, very cautious about reading too much into statements made by the latter, or for that matter into statements not made! As example, a few years ago I was told by a car salesman (from a very reputable dealership) that he understood from a company presentation that their future generation of cars would be powered by water, and was quite adamant about it. He then showed me a leaflet to back his point up - which showed what he was referring to was plans for HYDROGEN vehicles, with water as the only waste product. I'm sure that as far as his day to day job went he did a good job, and am 100% sure he didn't intend to mislead me - but sales people (even at quite a senior level) are not in general very technically minded.

In the case you refer to, it could be as simple as the marketing people in the consumer division see line-skipping as a big issue (because so many consumer DSLR type products have in the last decade) whereas in the pro division they don't. Pro consumers tend to have got used to pro levels of performance

Cliff Totten
July 8th, 2015, 02:00 PM
Yup. All goo points indeed.

At 2am last night, I was just about to hit the sack but decided to do another eye ball test using both cameras 4x focus expansion. A 30 second inspection seems to show the two very close when pointed at the fine detailed cloth textured sofa!

I do a semi scientific chart test on the weekend but I suspect it was a false alarm on my part.

I wasnt too serious or super confident about the topic in the first place.

CT :-)