ESTEBEVERDE
Veteran
Well... I hope all future mid range cameras have mounts for lenses so we can skip the whole adapter bit.
That is truly what I would like to see.
That is truly what I would like to see.
Why dont they FLIP???
- Thread starter Shane Jackson
- Start date
smelni
Veteran
there is a lot of confusing info in this thread.
THe hvx has a hardware switch to flip the image - I suggested months and months ago that they us the same function that the switch triggers to flip the image. But that function is not available to the firmware. In other words, it is likely that the physical switch trips a hardwired function within the LCD to flip the image - not a firmware function. So, as Jan said it is not possible.
In terms of PL mount and 1/3 chip - you will get the DOF of a 35mm lens with a PL mount regardless of chip size if you use one of 2 methods in an adapter.
- intermediate screen - this is the most common - by projecting an image on the intermediate screen the DOF of the lens is preserved for its full field of view thus allowing the camera, even with a 1/3 chip to record the image with the decreased DOF (more control over DOF, but wider FOV equals shallower DOF)
- using lens refraction - this is used, i beleive by the new JVC adapter. There is no intermediate screen but the light is bent after exiting the lens to "compress" onto the smaller chip thereby not cutting the image smaller (which increases DOF or makes it less shallow). This "compression" allows the sensor to see the full image.
If an adapter simply allows you to place a lens in front of the camera but doesnt do one of the above then you will not get the increased control over DOF ( or available shallower DOF). Your smaller 1/3 inch chip will just change the properties of the image to having that of deeper DOF.
I hope this makes some sense. - Optically I think the JVC adapter is ideal but that is why it is so darn expensive.
THe hvx has a hardware switch to flip the image - I suggested months and months ago that they us the same function that the switch triggers to flip the image. But that function is not available to the firmware. In other words, it is likely that the physical switch trips a hardwired function within the LCD to flip the image - not a firmware function. So, as Jan said it is not possible.
In terms of PL mount and 1/3 chip - you will get the DOF of a 35mm lens with a PL mount regardless of chip size if you use one of 2 methods in an adapter.
- intermediate screen - this is the most common - by projecting an image on the intermediate screen the DOF of the lens is preserved for its full field of view thus allowing the camera, even with a 1/3 chip to record the image with the decreased DOF (more control over DOF, but wider FOV equals shallower DOF)
- using lens refraction - this is used, i beleive by the new JVC adapter. There is no intermediate screen but the light is bent after exiting the lens to "compress" onto the smaller chip thereby not cutting the image smaller (which increases DOF or makes it less shallow). This "compression" allows the sensor to see the full image.
If an adapter simply allows you to place a lens in front of the camera but doesnt do one of the above then you will not get the increased control over DOF ( or available shallower DOF). Your smaller 1/3 inch chip will just change the properties of the image to having that of deeper DOF.
I hope this makes some sense. - Optically I think the JVC adapter is ideal but that is why it is so darn expensive.
Luis Caffesse
Admin
If I'm not mistaken the JVC adapter will give you DOF equivalent to 16mm film, not 35mm. Granted, that's probably enough for most - but its interesting to note.
I think it is a limitation of the lens relay system you were describing - because Angenieux has a similiar adapter (lens relay) which claims to keep the 35mm depth of field, but it is only made for 2/3" cameras.
Seeing the pricetag on those two adapters, I can only assume that either its not possible to create a lens relay to go from 35mm to 1/3" or (and more likely) it would simply be way too expensive to be a feasible solution.
Andrew McCarrick
Carbonite Member
I think the best thing I've seen (although it's not very easy; you have to make a custom mount.) is to mount the camera itself upside down and attach the 35mm adapter as if the camera were right side up. So the camera records the incoming image on the sensors upside-down from normal. Meaning it's recorded correctly .
Last edited:
Luis Caffesse
Admin
It can be even simpler than that too. As long as the camera is upside-down it shouldn't matter what orientation the adapter is. If you flip a projection screen upsidde down the image being projected on it is going to be the same orientation. So I think you should be able to mount the adapter in whatever orientation is ergonomically/practically best for your set-up.
davide
Andrew McCarrick
Carbonite Member
Very true.... didn't think of that.
Luis Caffesse
Admin
Are you guys being serious?
You are unwilling to deal with flipping a monitor upside down, or an upside down image in post - but somehow shooting with your camera upside down is perfectly acceptable?
I suppose different people draw the line in different places.... But I just can't imagine comfortablly shooting with my camera upside down for an entire shoot.
Regardless - its probably all a moot point given that ever major adapter has (or is about to have) an upright image. Finally!
You are unwilling to deal with flipping a monitor upside down, or an upside down image in post - but somehow shooting with your camera upside down is perfectly acceptable?
I suppose different people draw the line in different places.... But I just can't imagine comfortablly shooting with my camera upside down for an entire shoot.
Regardless - its probably all a moot point given that ever major adapter has (or is about to have) an upright image. Finally!
Dan Vance
DVXUser Sponsor
I would check with individual manufacturers of the adapters on that. Some of the spinning/oscillating mechanisms might not 'like' being operated upside down.
Dan Vance
DVXUser Sponsor
I just build all my sets upside-down and suspend my actors by their ankles from cables. So much easier than turning my monitor upside down--and risking dropping it in the process!
Luis Caffesse
Admin
Awesome.
DC
Cinematographiliac
Ok, since I'm the one who was going to start that original petition that Jan commented on, I'll chime in.
I would love to have the HVX200 flip its image like many other 35mm adapter users would. But it is just not possible to do this by a mere firmware upgrade. Sure, it sounds like it would be easy - just a few lines of code - I thought this too. There was a time when I knew the technical reason why. I'm not a video engineer so I don't remember exactly why. I will attempt to recall, as best I can, but let me be absolutely clear that the following info may be inaccurate.
The reason has something to do with the pixel sensors on the chips. There are areas where there are some faulty pixels on virtually every HVX200, as well as many HVX200's with a slightly misaligned chips to the lens (this is a big part of why the camera is priced accordingly). I believe the current design is such that faulty pixels are not recorded "as is" but rather interpreted by the nearest pixel sensors that are functioning. I think the detection of exact locations of faulty pixels are set at the factory for every individual HVX200. If the camera was to flip the image, the conflict comes into play when the physical chips are no longer oriented the way the software expects. I would then imagine you would get some nasties in your images. Panasonic doesn't want that (and neither should you).
Now, if anything I described above is proven wrong, I will gladly retract it or correct it.
Like Barry said, let's encourage Panasonic to implement the flip in future product designs. But also, the need for flipping is phasing out as 35mm adapters are incorporating their own solutions optically.
Last edited:
KyleProhaska
Veteran
Flipping in post just reverses the vertical pixel order....not much more to it than that.
- Kyle
ESTEBEVERDE
Veteran
Thanks so much for taking the time to reply DC.
Your explanation is both very interesting and at the same time a bit curious.
I agree with you up unto through the point where you say that flipping would destroy the relational database of real to mapped pixels.
But after that we diverge again.
The way an engineer explained it to me was that "I doubt image flip is hard - you just read out an image buffer in a different order."
It is clear that it is not the simplest of implementation to undertake for the HVX200 or Panasonic would most likely include it in a firmware upgrade.
It might very well be that the image buffers are hardwired and not reprogramable.
It would be very nice if the manufacturers could say weather or not their current models can be easily modified to to accommodate this or not.
According to you and some other reputable folks, Jan has spoken and said the HVX isn't.
That's very very cool that she is so responsive to not only Barry Green but the every day users here on DVX.
Perhaps when she has a moment she could have a knowledgeable product engineer give us brief explanation of the difficulties or as the case might very well be impossibilities of implementing this feature on the HVX.
And, yes, at this point the real solution is to have camera manufacturers include it in future models as well as making 35mm adapters themselves.
Who knows what lies ahead.
All the same, thanks so much for your time and follow up!
DC
Cinematographiliac
Sure, I understand. This information was relayed to me by a colleague (an HVX shooter who is also a good friend). And this colleague spoke directly with a Panasonic engineer about this. On a side note, this same engineer also thought it would be easy to implement, that is, until he looked deeper into it. It seems the HVX200's design just does not make the image flip-to-record a feasible feature through any kind of firmware upgrade.
Last edited:
There are a lot of assumptions and conjecture running around this thread, so I thought I'd chime in as well.
The HVX-200 is at it's core a computer, running a fully compiled version of Linux complete with specific device drivers for the components contained within. People on this thread are assuming that image processing has a lot of device driver and/or CPU action going on, and that there should be an opportunity to intercept the signal and 'flip' it. Knowing what I do about hardware I would assume that this is highly unlikely as signal processing is a form of streaming, and dedicated hardware is always faster than a CPU running software. It is much more likely that the devices within the camera are all tightly coupled and dedicated task processors operate with a minimum of CPU input. Anything beyond this and simple memory management may be beyond the capabilities of the CPU.
As an example, there is probably a digital signal processor chip that has the various gamma curves burned into it as binary mathematical equations. The CPU's only interaction with this chip would be to tell it which gamma curve is active for processing. The code burned into the chip does the rest.
ESTEBEVERDE, your 9 lines of code are actually mostly function calls, so what is the CPU overhead for the functions you are calling? The functions you are using will push data into a cache stack then pop them back out, First In Last Out. Does the CPU in the HVX have the clock speed to do this? Does it have the cache space? I believe it is moot because the data stream may not go through the processor that is running the Linux kernel. Even if it did the CPU would have to have access to enough cache to store 3 frames of 1080P material and enough bandwidth to stream it into cache and FILO it to the memory card before the next frame came along.
Some have commented that it must be possible because the image to the small LCD is flipped when rotated to the front. Flipping the image to the small screen is a trivial function of the GPU within the camera and probably works just like computer monitor rotation does. This is an interaction between a GPU and the LCD of a very reduced signal, which is a different animal altogether.
What I would like to see (which might be possible) would be to just turn off the image to the LCD while leaving the information overlay intact. That would make it easier to read while watching the image on an external monitor. A cool future product idea would be to put Histograms on the small LCD, but that would require additional processing overhead somewhere in the stream.
Finally, for those asking about the Hydra mod: you tether your camera to a computer which can do some post processing in the image it presents you (which is what is important). This is the same as the product that I use, DV Rack. DV Rack will not only flip the image while recording, it will also put it in a DV100 format which allows for true 1080 24P material which I can then load into a 24P timeline in Avid. Adobe bought DV Rack, now it is part of Premiere and is called On Location. If you are shooting a film with a decent budget it's the only way to fly...
Bob
The HVX-200 is at it's core a computer, running a fully compiled version of Linux complete with specific device drivers for the components contained within. People on this thread are assuming that image processing has a lot of device driver and/or CPU action going on, and that there should be an opportunity to intercept the signal and 'flip' it. Knowing what I do about hardware I would assume that this is highly unlikely as signal processing is a form of streaming, and dedicated hardware is always faster than a CPU running software. It is much more likely that the devices within the camera are all tightly coupled and dedicated task processors operate with a minimum of CPU input. Anything beyond this and simple memory management may be beyond the capabilities of the CPU.
As an example, there is probably a digital signal processor chip that has the various gamma curves burned into it as binary mathematical equations. The CPU's only interaction with this chip would be to tell it which gamma curve is active for processing. The code burned into the chip does the rest.
ESTEBEVERDE, your 9 lines of code are actually mostly function calls, so what is the CPU overhead for the functions you are calling? The functions you are using will push data into a cache stack then pop them back out, First In Last Out. Does the CPU in the HVX have the clock speed to do this? Does it have the cache space? I believe it is moot because the data stream may not go through the processor that is running the Linux kernel. Even if it did the CPU would have to have access to enough cache to store 3 frames of 1080P material and enough bandwidth to stream it into cache and FILO it to the memory card before the next frame came along.
Some have commented that it must be possible because the image to the small LCD is flipped when rotated to the front. Flipping the image to the small screen is a trivial function of the GPU within the camera and probably works just like computer monitor rotation does. This is an interaction between a GPU and the LCD of a very reduced signal, which is a different animal altogether.
What I would like to see (which might be possible) would be to just turn off the image to the LCD while leaving the information overlay intact. That would make it easier to read while watching the image on an external monitor. A cool future product idea would be to put Histograms on the small LCD, but that would require additional processing overhead somewhere in the stream.
Finally, for those asking about the Hydra mod: you tether your camera to a computer which can do some post processing in the image it presents you (which is what is important). This is the same as the product that I use, DV Rack. DV Rack will not only flip the image while recording, it will also put it in a DV100 format which allows for true 1080 24P material which I can then load into a 24P timeline in Avid. Adobe bought DV Rack, now it is part of Premiere and is called On Location. If you are shooting a film with a decent budget it's the only way to fly...
Bob