Improved designs, materials, and manufacturing could, I suppose, yield a sensor with higher resolution, greater light sensitivity, and lower noise. It all depends on how much Panasonic wants to advance the technology. I doubt that they will put that much effort into a prosumer AVCHD model.
Press Release: HMC150, "the new DVX"
- Thread starter Barry_Green
- Start date
Barry_Green
Moderator
Yes, the HMC150 is announced to be NTSC/PAL, 50/59.94 switchable.
Canon does it. I remember the hoopla caused by the A1's release because the image was dead on with the H1. The only difference was that the A1 had a fixed lens and that got people goin' "Well crap! I don't even change my lens! I'd rather use an adapter!"
A lot of us referred to Canon as CUT THROAT. But, I was happy that they did such a thing even though I never ended up with an A1.
So, maybe they would/could throw "theorized" new sensors into not only the HMC150 but say an HVX250? You wouldn't hurt either market because they would still BOTH very different machines. Just now when you say the HVX has about 8 stops of Dynamic Range, the noise floor (hypothetical situation involving advanced CCD technology) doesn't stop you at like 5 or 6 stops. =B That'd be nice.
Alex.Mitchell
Well-known member
If you're cutting long-GOP footage at anything other than an I frame it has to render that entire GOP as I frames, separate them, then recompress into I/P/B frames. At least that's how I understand it's done these days.
filmguy123
Veteran
I do NOT know this, but I doubt it is re-compressing the actual file to a file. I think how it works is, it decompresses on the fly so you can view it, marks the changes in the project file, and then actually re-compresses when you render. Anyone know for sure?
Digigenic
Well-known member
Well, in all fairness to Canon, their implementation of those CCDs in the H1, A1, and G1 was pretty much dictated by Sony, even though Canon maintained control of the way the video was processed and output.
But now, in addition to the ongoing "DIGIC" image processing developments, Canon is in fact pursuing the advancement and implementation of their own CMOS technology as evidenced by the HV20/HV30 etc.
As a result, Sony is becoming more aggressive in that department as well. The 1/2" Exmor CMOS chips in the EX-1 is a pretty clear indicator of that.
However, even though Panasonic also has access to in-house sensor development, with Matsushi*a being one of the top semiconductor manufacturers throughout the world, appears rather reluctant to take a leading position in that area, and instead are continuing the use of smaller sized CCDs while attempting to make up for it in other areas.
Panasonic took the same conservative approach with their DSLR system, by using a smaller 4/3 sensor system instead of the larger APS and/or full frame systems used by Canon and Nikon (who's current image sensor fabrication needs are being massaged by Sony).
So, depending on what we see announced by them at NAB, we'll know if Panasnoic's approach and/or perspective continues on the same conservative path in choosing to use only small CCDs, which I think it will, or if they're going to do something different.
BobDiaz
The Professor
To the best of my understanding, when doing a cuts only editing using any MPEG (like MPEG-2 or AVCHD) native editing, the in/out points are just pointers in the files.
Thus you can change the in/out point all over and not worry about re-rendering until the end.
HOWEVER, in the case of a title, color correction, dissolve, or some other effect, somewhere it has to be rendered. Thus if you want to see how it looks or maybe add something else to it, you will need to render it. The part that is rendered is only where the item is performed. So if a title is on 10 seconds of the clip, only 10 seconds has to be rendered and not the whole clip. (At least during the editing process.) In the end, everything needs to be rendered.
Having said that, remember each editing application is different, so it may not be exactly this way for the editing software you use.
Bob Diaz
Thus you can change the in/out point all over and not worry about re-rendering until the end.
HOWEVER, in the case of a title, color correction, dissolve, or some other effect, somewhere it has to be rendered. Thus if you want to see how it looks or maybe add something else to it, you will need to render it. The part that is rendered is only where the item is performed. So if a title is on 10 seconds of the clip, only 10 seconds has to be rendered and not the whole clip. (At least during the editing process.) In the end, everything needs to be rendered.
Having said that, remember each editing application is different, so it may not be exactly this way for the editing software you use.
Bob Diaz
filmguy123
Veteran
Yeah but when you render to preview an effect, it saves that to a different file, not overwriting the source. that's a non issue.
The sensor in the HV20/30 and Sony's Exmor chips manifest rolling-shutter artifacts (the former far more than the latter). I hope Panasonic doesn't migrate to CMOS sensors until it develops a design that is matches or exceeds its CCDs (e.g., global shutter, larger size, greater sensitivity and dynamic range).
BobDiaz
The Professor
After thinking about it, yes, you are right it should work that way. Best to avoid multi-generation de-compression/re-compression losses by always going back to the source file.
So editing in native mode may not be as bad as I first assumed. Other than any time delays for doing MPEG-4 AVC compression.
The justification for transcoding would be; after any delays for transcoding, rendering should be faster (in theory) and multi-layer work would have less multi-generation loss. The price one pays for this is more disk space for the transcoded files, time delays to transcode, and some generation losses due to transcoding.
A Minor Point On The Different Recording Rates:
The 6 Mbps, 9 Mbps, 13 Mbps, and ??? Mbps rates must change depending on the resolution and frame rate.
For example, assume that the highest rate is 20 Mbps. That will be OK for 1080@60i, 1080@30p, and 720@60p; however, drop down to 720@30p or 720@24p and the rate would be lower, just like on the HVX-200.
I'll guess that the camera will use terms like SHQ (Super High Quality), HQ (High Quality), SQ (Standard Quality), and EP (Extended Play) for the different data rates.
Bob Diaz
It's very unlikely that we'll see a global shutter CMOS chip. The quality of the image isn't as high and it's more expensive to make. That's why RED didn't go with a global shutter camera.
Digigenic
Well-known member
Well, it's not entirely clear to me whether or not RED had an active role in choosing one shutter type over another.
The option may not have even been available to them at the time.
Regardless, the wide-spread implementation of CMOS across all brands represents a significant shift within the industry, which Panasonic doesn't appear to be a part of.
While current trends in CMOS designs continue to show an equal number of strengths and weaknesses, it appears that the varations in artifacts between brands would signify that there are solutions being made to remedy the problems inherent in the earlier designs.
At this time, it really doesn't matter to me whether or not Panasonic implements CMOS.
What does matter though, is whether or not they are continuing to improve upon their current CCDs.
Under these circumstances, one has to wonder if there's anything left to gain from CCDs, and if there is, do those gains exceed what could potentially be gained from CMOS?
Based on the actions of the rest of the industry, it doesn't appear that there are.
BobDiaz
The Professor
More Thoughts... & Guesses
More Thoughts... & Guesses
Lens:
Anyone who has looked closely at the photo noticed that the wide angle of the zoom range said, "3.9mm"; however, the HVX-200 is 4.2mm. IF my calculations are correct, the lower mm of the lens will not get you to a 28mm equivalent, 30mm yes, but 28mm no. I'm going to guess that the CCD is roughly 16% more area than the HVX-200's CCD. That would give us the extra angle.
Now if all things are equal between the CCD of HVX-200 and the larger CCD of the HMC150, that would only give us less than 1/4 of a stop of gain with the extra area. However we are looking at about 3 years of incremental improvements in CCDs. I won't expect miracles, but some slight improvements could be expected.
While it's possible Panasonic has increased the resolution of the CCDs, you can't get something for nothing. Increased resolution results in lower sensitivity for the CCDs. So we can expect to either see some improvement in resolution OR some improvement in low light performance.
Here are possible choices: (1) Better light performance, but the CCDs are still 960h x 540v with both H & V pixel offset. (2) Improved resolution with CCDs of 1280h x 720v (with both H & V offset) OR 960h x 1080v with H pixel offset, OR ????? BUT low light performance suffers. I also am not doing CCD design, so I don't know how much they can increase the resolution before they get into problems. Remember the SSE (Split Screen Effect) with the JVC HD-100U? OK, JVC has solved the problem, but having to deal with a problem like that is not the place any manufacturer ever wants to be.
I think there are enough HVX-200 users out there who will stand up and say that the 960h x 540v CCDs of the HVX produce outstanding images. IF come NAB, Panasonic says it's native 960h x 540v, for the CCDs, as long as it produces HVX like images, I would be happy with that. My gut seems to say that the native resolution of the CCDs is going to be the same as the HVX-200.
On the other hand, if the pixel count goes up, that's OK too, as long as the image looks great, who cares what the paper specification says?
One more thing, if my guess is correct that the CCD is a bit larger, all thing being equal, the image should be able to resolve about 7% - 8% or so lines more than the HVX-200. HOWEVER, the big variable is the new lens, this will overshadow any slight changes in the CCD's size.
I hope the new lens turns out to be good, because I really like seeing a camera with a 28mm wide angle (35mm equivalent) at it's widest setting. Too many times, cameras like this are just not wide enough at the widest setting.
Bob Diaz
More Thoughts... & Guesses
Lens:
Anyone who has looked closely at the photo noticed that the wide angle of the zoom range said, "3.9mm"; however, the HVX-200 is 4.2mm. IF my calculations are correct, the lower mm of the lens will not get you to a 28mm equivalent, 30mm yes, but 28mm no. I'm going to guess that the CCD is roughly 16% more area than the HVX-200's CCD. That would give us the extra angle.
Now if all things are equal between the CCD of HVX-200 and the larger CCD of the HMC150, that would only give us less than 1/4 of a stop of gain with the extra area. However we are looking at about 3 years of incremental improvements in CCDs. I won't expect miracles, but some slight improvements could be expected.
While it's possible Panasonic has increased the resolution of the CCDs, you can't get something for nothing. Increased resolution results in lower sensitivity for the CCDs. So we can expect to either see some improvement in resolution OR some improvement in low light performance.
Here are possible choices: (1) Better light performance, but the CCDs are still 960h x 540v with both H & V pixel offset. (2) Improved resolution with CCDs of 1280h x 720v (with both H & V offset) OR 960h x 1080v with H pixel offset, OR ????? BUT low light performance suffers. I also am not doing CCD design, so I don't know how much they can increase the resolution before they get into problems. Remember the SSE (Split Screen Effect) with the JVC HD-100U? OK, JVC has solved the problem, but having to deal with a problem like that is not the place any manufacturer ever wants to be.
I think there are enough HVX-200 users out there who will stand up and say that the 960h x 540v CCDs of the HVX produce outstanding images. IF come NAB, Panasonic says it's native 960h x 540v, for the CCDs, as long as it produces HVX like images, I would be happy with that. My gut seems to say that the native resolution of the CCDs is going to be the same as the HVX-200.
On the other hand, if the pixel count goes up, that's OK too, as long as the image looks great, who cares what the paper specification says?
One more thing, if my guess is correct that the CCD is a bit larger, all thing being equal, the image should be able to resolve about 7% - 8% or so lines more than the HVX-200. HOWEVER, the big variable is the new lens, this will overshadow any slight changes in the CCD's size.
I hope the new lens turns out to be good, because I really like seeing a camera with a 28mm wide angle (35mm equivalent) at it's widest setting. Too many times, cameras like this are just not wide enough at the widest setting.
Bob Diaz
Elton
Veteran
I doubt you'd ever see CCD's with the res of the EX-1 in a Panasonic 1/3" camera.
I remember a thread back some time ago stating that a JVC engineer/rep said that their research showed that they ran into serious heat/noise issues with a 1/3" CCD beyond 720p resolution. My guess is that Panasonic wouldn't be able to sidestep the sheer physics of the problem in order to produce a quality full 1080p chip...particularly at 1/3".
The EX-1 obviously has some heat issues at its resolution. The damn thing has a vent near the rear of the camera to dissipate heat. You can see right through the slot! And that's a lower power CMOS chip.
I'm guessing the camera will have similar chips to the HVX.
On another note--if the manufacturers could run CMOS chips at 120hz (up to 120 fps internally), I bet that would negate a whole lot of the rolling shutter problems for most shooting scenarios. I believe that's what Grass Valley does in the new Infinity camcorder, but it has a much larger body for heat sinks, etc.
I remember a thread back some time ago stating that a JVC engineer/rep said that their research showed that they ran into serious heat/noise issues with a 1/3" CCD beyond 720p resolution. My guess is that Panasonic wouldn't be able to sidestep the sheer physics of the problem in order to produce a quality full 1080p chip...particularly at 1/3".
The EX-1 obviously has some heat issues at its resolution. The damn thing has a vent near the rear of the camera to dissipate heat. You can see right through the slot! And that's a lower power CMOS chip.
I'm guessing the camera will have similar chips to the HVX.
On another note--if the manufacturers could run CMOS chips at 120hz (up to 120 fps internally), I bet that would negate a whole lot of the rolling shutter problems for most shooting scenarios. I believe that's what Grass Valley does in the new Infinity camcorder, but it has a much larger body for heat sinks, etc.
Last edited:
Elton
Veteran
I don't think it's going out on a limb to say that.
filmguy123
Veteran
Umm. Equal strengths/weaknesses is subjective. CMOS is cheaper and prodcues less heat, and doesn't have smear problems Is there anything else going for it? There are a number of factors that are significant drawbacks, however.
Much of the industry has been adopting use of CMOS not because they are better, but because they are cheaper, consume less power/generate less heat. This makes everything cheaper. That is the benefit - NOT because it is a superior technology...
BobDiaz
The Professor
The SSE (Split Screen Effect) on the early JVC HD-100U was caused by the fact that in order to get all the data of the CCD quick enough, the CCD was split down the middle and had dual A/D converters, one for each side.
The problem was that getting the A/Ds to match exactly was a bit of a trick. When everything was cold, there was a mis-match that could be seen on a solid colored background. After about 10 minutes of "warming up" the mis-match was not as bad.
With a 960h x 540v CCD, a single A/D converter does the trick and your don't have a problem with an SSE. The Canon H1 was able to pull an interlaced 1440h x 540v in 1/60 of a second, so maybe it's possible to do 960h x 720v from a CCD without getting into trouble. I have no idea if they did make a 960h x 720v CCD, if the different scan rates are still possible.
I'll guess that the HMC150 will have the same 960h x 540v as the HVX-200.
As for future CMOS sensors, no question that lots of R&D will go into making them read faster. However, even if they could be read in 1/240th of a second, a random flash could still pose a problem, but would be less likely.
In the future, I'll guess the CMOS sensors will have an upper level and a lower level. On the upper level we'll have the light sensitive part and on the lower level will be the part that shifts the data to the outside world, sort of like a subway system under the city.
Thus, the image would be exposed, the voltages drop down to the lower level through a transistor switch, the switch closes, and the data is shifted out at normal scan rates.
Bob Diaz
Digigenic
Well-known member
Yes, it does have other things going for it.
Aside from the benefits you've already noted, CMOS designs fascilitate the use of smaller circutry on chip and larger/gapless and layerable microlens designs.
These examples can been seen in ongoing developments made by Canon, Nikon/Sony, Fuji, and Sigma/Fovean.
I never said CMOS was superior.
I simply suggested that CMOS held more promise.
The R&D these other companies have been pouring into it should be indicative of that.
By being cheaper, consuming less power, generating less heat, while producing images that to many eyes are as good or better than images rendered from CCDs then that by some persons'/corporations' definition does make it superior technology.