Is the 2k raw in 7Q going to be as bad as 2k in sonys raw recorder?

S

Sami2

Active member
Hi
here is presentation about the 4k with sony recorder:
http://vimeo.com/69884277

As you can see from the tests they made with 2K raw the results were horrible compared to 4k hence 2k raw is already prosessed internally to half the resolution.

Since i was unable to find any sample clips from 2k out of Q7 I have to assume it is the same stuff as with sonys recorder, has anyone have any news on this?

If this really is the case it seals the deal of going with sonys recording option and getting the 4k raw with a "nice" price tag...
 
Well, raw is raw. So how could the 7Q record the raw stream any differently than Sony's recorder? The 7Q can only record what it's given, so the quality of the raw that the 7Q records will be exactly what Sony feeds them, nothing more, nothing less.

As to the quality of the raw that the camera outputs -- there's really only two ways to get 2K off of a 4K sensor. You either do it Red's way, windowing the sensor down and extracting a smaller portion of the sensor at its full 100% raw, or you do it by skipping pixels or processing it somehow, to retain the full field of view while simultaneously discarding 75% of the data (2K is 1/4 the data of 4K). So I wouldn't be surprised if the 2K quality were lower than the 4K, I'd actually expect that.

They can't scale it down in-camera because it's raw, they're presenting raw data. If they're digitally scaling it, that would mean a substantial amount of processing was going on with the signal and it was no longer raw.

This is all just general guesses though, I don't know what Sony's doing in-camera.
 
To say that the 2K is "horrible" is quite a leap, in my opinion. Even the presenters say that it is "merely very good." I wouldn't say that merely very good = horrible. It is true that the 2K coming out of the FS700 will have less clarity than the 4K output, but that doesn't mean at all that it's going to look horrible.

They're only saying that the 4K is outstandingly sharp and that 2K doesn't exactly reach that level - but is still very good.

Unless your video work consists of shooting lots of detailed video charts like the ones they show, I wouldn't be toooo concerned and I'm personally waiting until actual 2K footage from the FS700 hits the web to make my judgement. Still the 7Q can record the 4K signal (non-RAW) up to 30 fps, so the 7Q will get some access to the super-crisp 4K imagery.
 
I think it is pretty fair to say that 2k raw is horrible considering that it has the same resolution as the internal 240 fps at the moment which has its resolution below HD. So considering that when recording 2k raw you are actually getting less resolution than when recording fullHD avchd internally you gotta be dissapointed. 2k raw done at 2k resolution would have satisfied my needs just fine but now i have to consider getting the 4k

Raw is always not raw sony here demonstrates by doing some debayering or other tricks before hand inorder to get the signal down to 2k. Sony does not want to tell what they are actually doing to get 2k.

Also Canons 5d mkIII magic lantern raw hd is not raw hence it has totally different sensor resolution.
 
I watched their entire presentation earlier this morning and have seen the segment you are referring to. I'm not saying that it's not a letdown to hear about this loss of resolution in the 2K stream... I was just pointing out that Adam and Art did not go near declaring it "horrible."

I've not had much trouble using the 240fps footage in my productions and haven't seen any terrible aliasing or moire issues in the footage I've shot, so I'm just still curious as to how much of a real-world issue the 2K image is going to produce. I'm anxious to see more tests and sample footage from the FS700's 2K signal before getting too discouraged by the charts they showed in the presentation.

Also, the 2K stream is still RAW in the manner that it is formatted as raw sensor data and still has all the capabilities of RAW in post. But as Barry stated earlier, to get a 2K image out of a 4K native sensor, you have to do some sort of processing. In this case, Sony is merging the sensor data from nearby sensor areas to create fewer pixels and therefore, a smaller image. It is this process that is most likely causing the difference in resolution in the 2K signal. The 2K stream is not debayered and is still technically RAW.
 
inorder to merge pixels they have to do some debayering and then spin the footage backwards to make it look like raw again.

You could say that ff course those guys are not saying it is horrible hence it is a presentation supported by sony.
There was the potential of 2k being super good, even better than 4k at pixel level since there is more data for calculating the color information per pixel than in 4k coming from sensor when you do 2k.

I consider everything below the original capabilities of the camera a step backward and that is what the 2k raw at normal frame rates at the moment is in terms of resolution. So i will rather get the sonys recorder and get the super crispy 4k raw 60fps instead of the 7Q with a low resolution 2k "raw". why not just save some space and make it 720p raw instead of 2k hence that is about the resolution it really is
 
Or they could actually just merge pixels with some weird algorithm and then do the magic again in raw viewer. Something they do not want to tell the public what they are actually doing.
 
from what I understand, gang, it's not some dark voodoo or great mystery... Sony simply employs pixel binning when getting a fast read (SSM) or 2K off the sensor. it's proprietary in that they have come up with their own algorithm to average the reads of blocks of pixels, combining them into 1 block, but it's hardly new or novel tech. it's the same tech (more or less) that accomplishes the scaling of pictures in some photo editing software. the problem is, to do it really well requires a lot of processing power... something the camera clearly lacks. but it IS noticeable (outside of focus charts) in the way it reduces resolution and, in my experience, also results in a subtle color shift between 24p rendering of a scene and 120 or 240... so it IS concerning that we could be seeing those same artifacts in all of our 2K material. but to say it will be horrible (or even bad) is too soon. it will be 12 bit raw, which is a BIG step up from AVCHD or 8bit external recording. honestly, my hope is that CD can find a way to scale the 4K raw to 2K on the 7Q (which has huge processing power) in order to derive a cleaner 2K. but if not, we'll have to accept that it's great at what it was designed to do: 1080p & 4K, and good at, what is arguably a bonus feature: 2K.
 
blackdogdylan said:
from what I understand, gang, it's not some dark voodoo or great mystery... Sony simply employs pixel binning when getting a fast read (SSM) or 2K off the sensor. it's proprietary in that they have come up with their own algorithm to average the reads of blocks of pixels, combining them into 1 block, but it's hardly new or novel tech. it's the same tech (more or less) that accomplishes the scaling of pictures in some photo editing software. the problem is, to do it really well requires a lot of processing power... something the camera clearly lacks. but it IS noticeable (outside of focus charts) in the way it reduces resolution and, in my experience, also results in a subtle color shift between 24p rendering of a scene and 120 or 240... so it IS concerning that we could be seeing those same artifacts in all of our 2K material. but to say it will be horrible (or even bad) is too soon. it will be 12 bit raw, which is a BIG step up from AVCHD or 8bit external recording. honestly, my hope is that CD can find a way to scale the 4K raw to 2K on the 7Q (which has huge processing power) in order to derive a cleaner 2K. but if not, we'll have to accept that it's great at what it was designed to do: 1080p & 4K, and good at, what is arguably a bonus feature: 2K.
Click to expand...

I don't know if they are pixel binning or not. It sounds like they are, but colour pixel binning results in a single RGB pixel that can't be debayered, because there is no bayer pattern anymore. They have to be doing something different. I wonder if they are pixel binning, and then splitting that single pixel back into a bayer pattern. It's also possible they could be line skipping.
 
From the looks of the resolution in 2k they could even be doing line skipping by taking every third linen from the sensor hence taking every second drops out totally red or blue value.
You cannot just bind tthe pixels together, you have to now to which color each pixels luminance corresponds to since you are only recording the luminance in raw.

It is totally fair to say it is bad when taking into account only the resolution, since by paying 4k extra you get less than you started with.

No wonder there are not any sample clips out of 7Q, who knows what this process will do to the raw color information.

I think this is an issue that needs to be adressed and clearly it raises the question of is sony going to fix this in future firmware updates or are we just stuck with brilliant colors, headroom ja low res as the result what the 2k raw will provide us.

4k uncompressed in 7Q (is it going to be 422 who knows) is a nice option but in sony it is freaking 4k RAW
 
Sami2 said:
From the looks of the resolution in 2k they could even be doing line skipping by taking every third linen from the sensor hence taking every second drops out totally red or blue value.
You cannot just bind tthe pixels together, you have to now to which color each pixels luminance corresponds to since you are only recording the luminance in raw.

It is totally fair to say it is bad when taking into account only the resolution, since by paying 4k extra you get less than you started with.

No wonder there are not any sample clips out of 7Q, who knows what this process will do to the raw color information.

I think this is an issue that needs to be adressed and clearly it raises the question of is sony going to fix this in future firmware updates or are we just stuck with brilliant colors, headroom ja low res as the result what the 2k raw will provide us.

4k uncompressed in 7Q (is it going to be 422 who knows) is a nice option but in sony it is freaking 4k RAW
Click to expand...

You can bin pixels. In fact, that is how the Canon C100 and C300 work, they both have QuadHD image sensors at 3840x2160, and do a 2x2 binning down to 1920x1080. So they grab the RGBG block of 2x2 and turn it into 1 RGB pixel.

I don't believe that is what the FS700 is doing (but they should if they aren't). If they are line skipping, they will be doing it both horizontally and vertically, because they are going from a 4096x2160 pixel area down to 2048x1080, so they need to halve both vert and horz. That means they are probably skipping 2 lines in each direction. Effectively giving you a 2048x1080 image sensor, with 2px wide black lines between each pixel group.
 
as you make it to one RGB pixel it is not raw anymore.
Line skipping cannot not be done by just taking every second line due to bayer 2x2 structure hence you will loose either blue or red.
skipping three line will result in 1366x720 that is not too far from what the 2k looks like at the moment.

Canon always does something differently hence they do still photography that is why i chose sony instead of canon. And i am very happy that sony is taking a different approach on these camcorders so we can enjoy shooting nice images :)
 
Sami2 said:
as you make it to one RGB pixel it is not raw anymore.
Line skipping cannot not be done by just taking every second line due to bayer 2x2 structure hence you will loose either blue or red.
skipping three line will result in 1366x720 that is not too far from what the 2k looks like at the moment.

Canon always does something differently hence they do still photography that is why i chose sony instead of canon. And i am very happy that sony is taking a different approach on these camcorders so we can enjoy shooting nice images :)
Click to expand...

It wouldn't be every 2nd line. It would be every 3rd and 4th line. Also, Sony does still photography too, and Canon makes amazing video images as well.
 
i meant to say taking every third line :D
I have not soon anything amazing from canon under 20k and they are the company to blaim for ruining the video production ergonomics by making it possible to record bad video on 5d markII. By bad i mean no DR, maybe six stops in real life situations and no gradients on skin tones etc. The lack of gradients in skin tone is also a problem still photography equipment, but they have no reason to fix since people are already buying that crap anyway. Let the money flow into the company and the R&D can rest 20 more years
Sony instead is taking a totally different approach and boldly created the monstrous f65 and is deriving stuff downwards from that.

The reason why i disrespect Canon is that it has been making digital cameras for around 20 years now and we have not seen practically any significant increases in DR from them in the past 10 years or so. However the resolutions have gone up to ridiculous measures for 35mm sensor. They are just making stuff quaranties the cash flow and are not pushing the boundaries. Thahk god Nikon started makin great DSLR systems and has gone way beyound canon on image quality etc.

Well you can always call lack of dynamic range and skin tones amazing. I Would not that is why is chose sony how. Before that the brand to go with was panasonic and their p2
 
Unless you're going above 120 FPS then the CD 7Q should be able to scale the 4k raw to 2k Prores 444 in 12 bit. It should be able to scale the 4k raw to 1080p Prores 422 in 10 bit. It should be able to record 4 raw as 4k prores 444 in 12 bit. These options are more than enough for feature film production in my opinion. They are certainly more than enough for broadcast work. To me these are more attractive options than raw.

Please let us know Dan if CD are not planning this implementation.
 
It makes me wonder if CD is just a big hoax since we have not seen any sample clips from their 7Q. Correct me pease if there are some available.
Usually there is plenty of samples from preproduction models from every manufacturer especially when the final product is supposed to hit the stores in next months.

"Unless you're going above 120 FPS then the CD 7Q should be able to scale the 4k raw to 2k Prores 444 in 12 bit."
Will sony allow this hence they are at the moment allowed to get 30fps in 4k, so taking 60fps and downscaling it down to 2k in 7Q might be something that sony will not allow them to do since sony is already supplying them 2k directly from camera.
 
atarijedi said:
I don't know if they are pixel binning or not. It sounds like they are, but colour pixel binning results in a single RGB pixel that can't be debayered, because there is no bayer pattern anymore. They have to be doing something different. I wonder if they are pixel binning, and then splitting that single pixel back into a bayer pattern. It's also possible they could be line skipping.
Click to expand...
"Pixel binning" means something very specific - adding together the charge before A_D conversion from more than one photosite. It's quite possible within a largish block to separately bin the red, green and blue sites to end up with three single values for each colour.

In this case, the geometry means they must be doing something that halves resolution symmetrically in the H&V directions, and bearing in mind the 2x2 nature of Bayer, the simplest method would be as you say - ignore all photosites in every 3rd and 4th line, and/or every 3rd or 4th column. As you say: "Effectively giving you a 2048x1080 image sensor, with 2px wide black lines between each pixel group."

There are variations on it, but they all involve deriving a 2x2 Bayer pattern from a 4x4 section of the chip. (Another possibility would be to treat the 4x4 block as 4 2x2 blocks, then use the red photosite from one block, blue from another, and two greens from the other two blocks.)

One thing that hasn't been mentioned is that even allowing for this, the result of the chart posted is worse than would be expected, and I wonder if what we're seeing may be partly due to too simple "deBayering" of the 2K RAW? Specifically, what has been shown is not the result of a true deBayer, but looks like what would be expected from simply deriving an R,G,B pixel from each block. (As the Canon C300 does.) That would give half the inherent chip resolution (as we see), whereas a conventional deBayer should give about 80% for luminance (about 860 lpph).

Is it possible that what has been posted is just intended for viewing - and the actual RAW file will give a far better result when properly deBayered?
 
blackdogdylan said:
from what I understand, gang, it's not some dark voodoo or great mystery... Sony simply employs pixel binning when getting a fast read (SSM) or 2K off the sensor. it's proprietary in that they have come up with their own algorithm to average the reads of blocks of pixels, combining them into 1 block, but it's hardly new or novel tech. it's the same tech (more or less) that accomplishes the scaling of pictures in some photo editing software. the problem is, to do it really well requires a lot of processing power... something the camera clearly lacks. but it IS noticeable (outside of focus charts) in the way it reduces resolution and, in my experience, also results in a subtle color shift between 24p rendering of a scene and 120 or 240... so it IS concerning that we could be seeing those same artifacts in all of our 2K material. but to say it will be horrible (or even bad) is too soon. it will be 12 bit raw, which is a BIG step up from AVCHD or 8bit external recording. honestly, my hope is that CD can find a way to scale the 4K raw to 2K on the 7Q (which has huge processing power) in order to derive a cleaner 2K. but if not, we'll have to accept that it's great at what it was designed to do: 1080p & 4K, and good at, what is arguably a bonus feature: 2K.
Click to expand...

+1.

2K is the bonus feature they added just before the NAB, 4K is the stuff they promised us one and half year ago.

i've checked FS700's 4K stuff at Sony's private exhibition last month and it was superb. i saw it with their 4K master monitor and it blew me away totally.

FS700 will give us crazy sharp 4K images from 24fps to 120fps(4 sec burst) without any resolution loss. this is such a unique feature.


freiheit
 
You must log in or register to reply here.
Back
Top