Does the FX7 need a lot of light or what?

[Barry Green]

Originally Posted by Tony Tremble

Why is it the HVX is faster than the Z1 yet the Z1 is a superior low light performing camera?

I don't accept that it is. I accept that it's cleaner, yes, and can be "gained up" to deliver comparable sensitivity. But I don't like how the color drains out of the picture on the Z1 when you do that.

While I'm getting the third degree, let me ask you why you proclaim that the Z1 is "a superior low light performing camera?" Do you have methodology and datum to support that position, or are you (to use your word) "regurgitating" what someone else has said about it? I have tested them both, with 11 other people in the room including Nate Weaver, Adam Wilt, Shannon Rawls, Jay Nemeth, Aaron Umetani, Evin Grant, Rush Hamden, and a few others, and our ISO determinations were that the HVX is 320 and the Z1 is 160.

But that isn't a determination of "low light" superiority, that's sensitivity.

Originally Posted by Tony Tremble

What is the absolute value by which you measure a camera's low light capability?

I don't know that anybody has, or that they definitively can. We've measured ISO sensitivity at 0dB gain, but that is not a "low light capability" test, it's a sensitivity test. A comprehensive "low light" test would be quite subjective and have to take into account someone's tolerance threshhold for noise, the particular type of noise and their relative acceptance of that type of noise, some way to try to equalize noise at gain (as a quick eyeball, a Z1 at 9dB is about equiv. to an HVX at 0dB noise-wise, but how do you specifically quantify that?)

So I was quoting sensitivity measurements, not "low light" measurements.

Why is the HVX more noisy than the Z1 or canon A1?

Because it's progressive and they're interlaced. Under otherwise identical circumstances you can expect a progressive chip to be 6dB noisier, and 6dB brighter, than the same chip scanned in an interlaced fashion (which is easy enough to do, just try it on an XL2 or DVX, both of which scan progressively and interlaced). If you have 320 ISO from a progressive chip, that stands to reason that it would be noisier than 160 ISO from an interlaced chip. The interlaced chip can be tuned much "quieter" because it'll benefit from a) the sensitivity boost garnered from scanning interlaced, and b) the noise reduction effect of row-pair summation.

And the A1 is far noisier than the Z1, most likely due to it having substantially smaller pixels than the Z1. The A1 is about equally noisy with the HVX if you don't engage the NR1 function in the A1, but the noise is of a different texture; the A1's noise is grainy/bitty luma noise, the HVX's noise is splotchy/smeary chroma noise. They appear to be about the same sensitivity, with the A1 enjoying a maybe 1/4-stop advantage in raw sensitivity.

Engaging the NR1 function does help clean up the noise in an A1's signal, but I don't know what penalty it involves. There's always a tradeoff in any engineering decision, and there's a reason that NR1 isn't engaged by default. But I don't know what that reason is. There must be some negative effect to the image or else they would have made it the standard; they've said that some ghosting can happen under other noise-reduction settings, but there's got to be some drawback to engaging NR1. Until I find it, I continue to use NR1 and it does help control the noise.

How are you measuring ISO and how are you normalising the measurements.

When I measure ISO it's by using an 18% gray card and a waveform monitor, under a constant level of illumination, tuning the iris to deliver 55 IRE on the waveform monitor, and then cross-referencing that with a spotmeter, and adjusting the ISO on the spotmeter until it reads the same iris setting as the camera. Furthermore, knowing that cameras have finer increments than the f-stop readout can display, I will fine-tune the light until the camera's iris is exactly registering just over the "border" to the next appropriate full f-stop before taking a reading with the spotmeter.

Are you simply regurgitating someone else's pontification without knowing how they conducted the measurements?

Your choice of the words "regurgitating" and "pontification" speaks volumes about your attitude in this discussion. But yes, as pointed out in the very first post, I was "regurgitating" Adam's results, since those are the only ones available to us. And far from "pontification", he's a brilliant Princeton-trained engineer who knows how to properly conduct tests. He's the only one who has tested the cameras for sensitivity, other than Jon Fordham saying that the V1 was "about a stop slower than the Z1," a statement which would support Adam's findings even if not scientifically conducted. And having worked with Adam side by side on the six-camera shootout, I believe I know his measurement methodology plenty well and trust that he would take adequately scientific measurements to determine ISO. Hardly "pontification."

If that isn't good enough for you, please feel free to conduct your own test.

[Marcus Marchesseault]

All flames aside...

I find Brian Rhodes FX7 clips to be very helpful. In fact, I think it may partly disprove that the FX7 is slower than the Z1/FX1. I find the images in low light (although noisy in auto settings) to be more appealing than those from the FX1. I think it is because the color seems to stay correct. I am not bothered by the noise nearly as much as I am bothered by a camera losing it's color in low light. Even my very sensitive VX2000 would start losing color at 12db. I think Spot was right on the money when he stated that the appearance of the video to the viewer is more important than specifications or resolution charts. I find the V1/FX7 images to be more appealing than the Z1/FX1.

I also think the manual settings in the night traffic scene had plenty of exposure. Yes, there are some bright signs illuminating the area, but everything looked like it should to my eyes. I think exposure could even be dropped slightly. What I really like is that all the bright lights weren't bleeding out the entire scene. A CCD camera doing that same shot probably wouldn't have left the signs readable and the cars would be very obscured by their own headlights. I think this may make the FX7/V1 even more functional in night scenes than other cameras. Remember, in a night scene it is the lighting that is important. Anything in a night scene not lit doesn't exist. If the FX7/V1 handle lighting better, they are functionally better than a CCD camera in many instances.

As long as the FX7 can expose nighttime lighting without using a lot of gain, it is a perfectly useable low-light camera. The only instance where light goes lower than a cityscape is in a dim ballroom during something like a wedding. Event videographers accustomed to the PD/VX cameras will need to use supplemental light to work with the FX7.

A question for Brian: Do you remember the manual exposure settings you used for the night traffic scene? How much gain, if any, was used? Was the shutter at the default 1/60th?

[Douglas Spotted Eagle]

Marcus, as much as I appreciate and enjoy the V1, it's slower than the Z1, no question. Is it a full stop slower? I don't believe so. I didn't conduct sensitivity tests on the preproduction model, because I knew that could potentially change, and also elicit this sort of debate. I did conduct resolution tests because I was specifically asked to.
that said, I also am of the opinion that the V1 (and likely FX7) are "smarter" than the earlier generation HDV camcorders, and that's the reason why we see the 'clean' images. Tremendous labor has gone into the DSP on these cams, and it shows, IMO.

[Marcus Marchesseault]

I'm sorry for the long post, but I think I am on to something important regarding light handling characteristics of different imaging devices. I bring up a few good examples, but I know people get bored with long posts. My summary is this: "If the V1/FX7 handle lights better, they handle the movie set and stage better." If you are in the market for a new camera, the characteristics I describe may have you thinking about these cameras depending on your shooting conditions.

*******************************

"I also am of the opinion that the V1 (and likely FX7) are "smarter" than the earlier generation HDV camcorders, and that's the reason why we see the 'clean' images. Tremendous labor has gone into the DSP on these cams, and it shows, IMO."

Spot, I think that is the technical reason why I like the V1 images (thanks for all of your and other's work to bring images out), but I don't think the whole story is in the tests or the numbers. One of your articles states that CMOS chips in the V1 have individual addresses. That is a big technical detail that I think Sony has capitalized on. I mentioned that you have said that the perception is even more important than the tests and I wholeheartedly agree. One reason I agree is that I try to imagine the audience's perception. Will they care about the technical capabilities of any particular camera? No, they will judged by the conscious and subconscious of the aesthetic. What I am guessing about the V1 is that it will be easier to get aesthetically pleasing images in most situations due to it's inherent differences in the imaging device. The exception will be very dimly lit rooms like ballrooms. What I don't think will be a limitation will be normally lit scenes. I like the images I have seen from the FX7/V1 because the color is retained. Yes, resolution is important, but I think the better color and handling of highlights makes things seem better. A bit of noise is not a problem for me as long as it isn't accompanied by The Horrible Yellow Tint. There are some great ideas being discussed over in the wedding and event videographer forums about how to light a ballroom without destroying the dim-lighting aesthetic. With a solution for low-light in ballrooms being provided by other users of this forum, this leaves all of the other situations to consider.

Here is where I think these cameras will excell:

1. Outdoor daytime light. I can't tell you how much I would love to actually have blue sky without making everything in the foreground go to silhouette. Even if the sky does overexpose, it will be easy to tint it a bit blue because of the relative lack of smear. Spot, your shots of the open parachute passing by bright areas of sky convinced me of this. Those parachute lines are tiny and I know they would have disappeared on my old VX2000. I suspect the Z1 also couldn't deal with lines that small in front of a bright sky. I really like both the VX and FX1, but I think the V1 is even better because I really want to get nice shots outdoors.

2. Artificially lit night scenes. Is there any other kind? Unless a camera comes out that can get proper exposure with only moonlight, low light situations are all artificially lit. By not smearing light sources all over the scene, the FX7 keeps things in the cityscape looking like they should. This should make utilizing practical lights even easier now that they will look right. I use the night traffic clip from Brian Rhodes as an example again. Look at the headlights on the cars. Yes, they have lens flare, but you can still see a bit of the front of the car and all of the rest of the car as it passes by a light source.

What I think most people may have missed is that you can still see the color of the headlights. There is still color information being preserved in a car headlight! That is a huge deal. This means that FX7/V1 users can have lights in the scene of fairly strong intensity and keep them looking natural. Just read the big signs in the night driving shot. I know that wouldn't work with most video cameras. If lit signs all turn into big blobs, that is a good way to kill your illusion. A good example of the benefits for the movie maker would be replacing the bulb in a practical lamp with something twice as bright (a big compact fluorescent, for instance) and it will still look like a regular lamp. If you put a bright light in your scene with many other camcorders, it will blow out and smear all over the scene. The detail of the light, and much of what surrounds it, is lost. Now, with the V1/FX7, practical lights can be pumped up to give a room enough baseline exposure so that the hot studio lights won't need to do all of the work. This could easily lead to a more natural look in low-budget productions. There are other implications to better highlight handling and lack of smear, but we will have to learn them once the cameras are in our hands. I think people will be surprised what difference this one seemingly small detail will make.

Whatever is not lit, doesn't exist in the final production. If a camera can make it easier to light a scene, it is functionally better in many low-light conditions. Artificial lighting is all about contrast ratios, not about the actual quantity of light. A brightly-lit room can look very dark on-camera if the shadows and black spaces are correct.

I JUST THOUGHT OF A GREAT EXAMPLE! Picture Douglas Spotted Eagle onstage playing a flute. It is a night scene or indoors. With what will he be lit? A spotlight. Spot in the spotlight. How will a camera like the V1 have an advantage? If it has greater latitude and can deal with highlights better, Spot will not turn into a big white blob in the middle of a black background. Instead, he will still have detail on his face. His fancy clothes will still be visible. He will still look like a person in front of a dark background instead of a glowing white disembodied torso (spotlights don't always cover the whole performer). Any lights in the area will not look like disembodied blobs of white, but will retain some of thier original color and shape. If the background has some light on it to create a silhouette effect on mid-ground objects/subjects (like backing musicians), it will retain it's color instead of bleeding out the silhouettes and turning to white. Yes, a fair amount of light is being used, but the scene will still have that overall dark look due to so much area onstage being unlit.

If the V1/FX7 handle lights better, they handle the movie set and stage better.

[Tony Tremble]

Barry,

Sorry for my poor choice of words and thank you for the trouble you've taken in you reply.

I don't understand why sensitivity is not directly related to low light performance. Sensitivity should be directly related to low light performance. What am I not understanding?

The tests appear to favour a noisy camera over a less noisy camera even though the less noisy camera's gain could be increased by 9dB (as you note) to match the noise on both cameras. This then would be a reference by which the cameras could be compared. I appreciate the subjective nature of comparing the noise but that has to be more accurate than ignoring altogether.??

I am not trying to be belligerent or catch you out in any way I am just trying to understand how the values compare. The problem is I can't see the datum by which the cameras are being compared. Setting 0dB on each camera doesn't seem a valid reference point and thus the ISO values are an unreliable measure of actual sensitivity.

If by scanning progressively a camera gains a 6dB boost then why an interlaced camera shouldn't have 6 dB dialled in to make the test comparable. This would also suggest that different ISO values would be recorded depending on whether the Sony V1 was measured in progressive mode or interlaced. In progressive it would claw back its one stop to equal the Z1. Is that a fair assumption?

Cheers
TT

[Steve Mullen]

[QUOTE=Marcus MarchesseaultIf the V1/FX7 handle lights better, they handle the movie set and stage better.[/QUOTE]

Everything you've said matches my experience with the V1. I keep trying to create a Japanese type ad tag:

"... captures what your eyes see."

"... captures as your eyes see."

In about 10-days I'll be stating a series for camcorderinfo.com I call "Postcards from the *cutting* Edge -- the V1 in India." It will have some screen shots that show that even at night -- the scene it captures matches what you see --especially in terms of color and color of very bright objects.

Sensitivity numbers are really not going to be of universal value given CMOS and DSP. The key is to know WHEN to use gain and HOW much. In many situations 15dB produces noise, but since the picture still looks perfect "other" than the noise -- one's eyes simply ignore the noise and accepts the picture. No different than Discovery HD -- once accepts that in a dark cave there will be mild video noise.

In other situations I would not use more than 6-9dB and let the pix be dark and clean.

I'm not saying I wouldn't like 6-12dB more sensitivity, but I expect the Z2 will bring this next year.

[Tony Tremble]

Originally Posted by Steve Mullen

In about 10-days I'll be stating a series for camcorderinfo.com I call "Postcards from the *cutting* Edge -- the V1 in India." It will have some screen shots that show that even at night -- the scene it captures matches what you see --especially in terms of color and color of very bright objects.

Two reasons for me to be green with envy, you have a V1 AND you are in India. I've have the good fortune to travel extensively through India but only with my trusty PD150. I try and go for 3 months every year to lose myself in the colour and the smell. A perfect antidote for the western way of living.

Steve what is your itinerary?

Looking forward to some wonderful HD shots of Mother India.

TT

p.s Looking forward to doing it all again but with an upgraded camera!!! :)

[Thomas Smet]

Marcus,

That night shot with the cars has to be one of the ugliest looking HDV shots I have ever seen from any of the HDV cameras. I have never seen so much CA in one single shot in my entire life.

I'm not sure where you are coming from but the headlights look like garbage to me. They all have blue halos. Also look at the streetlights. The amount of blue and purple CA is in my opinion totally useless.

I realize this may not be the best piece of example footage but you are using this piece of video to talk about how good the camera is? The amount of CA in this shot is more extreme then some of the first night shots from the JVC HD100.

[Marcus Marchesseault]

First, a clarification. I'm not referring to the purple fringing as color detail in the night driving scene. I should have been more clear, but I thought I was being too long-winded already. I was referring to the fact that the color temperature of the lights from the different vehicles is still detectable. I can tell the daylight-balanced lights apart from the tungsten/halogen headlights. The effect many CCD cameras will have in the same situation is to bleed the headlights all over the place and have the color go to pure white.

On to Chromatic Aberration: No, it is not ideal to have the dreaded purple fringe. Unfortunately, this is a function of physics and I see it in many cameras. I believe the brighter the source, the more apparent the CA. Why is this not a problem for the V1 in my opinion? I see two reasons. Most people won't care when they see the video. Luckily, the CA is only around the really intense sources and not really visible around the lighted signs. The fact that there is anything visible around those headlights and streetlights at all is a miracle of modern science. Previous cameras wouldn't show CA because it would have been smeared out by the CCD block. The best thing is that the purple fringe is a very distinct color, so it can probable be keyed and have it's color adjusted. The fact that the strong light colors are still retained means you can probably tint the purple fringe to the color of the actual light in the scene.

The way to deal with the purple fringe (which only seems to show up around an intense light source) is going to be similar to the method to fix overexposed sky. Make a luminance key and add a tiny hint of blue. If the sky doesn't bleed around the subject, it can be keyed and adjusted. Color correction is our friend. I think it's time for me to get a tutorial DVD on color correction. I haven't done these techniques myself, but other people have that visit this forum.

Of course, this kind of detailed color correction won't be practical on all types of productions, but at least there will be a possible way to fix the issue. I really don't think the audience will care about the purple fringe in most cases. In scenes where it becomes really noticeable, be thankful that there is only a bit of purple and not a field of ghostly white tracking your light source.

I really think the great news is that the V1 offers some control where there once was literally chaos. The apparent exposure latitude and relative lack of smear will bring the artifacts/errors down to a level where a good technician will be able to pull out detail in the highs and lows as well as correct colors to what is perceived by a human eye. I guess what I am saying is that the faults of the V1 are something that we can deal with. Before, there was nothing to be done about overexposure issues. Now, there will be just enough information left in the highlights that we can grab ahold of that detail and use it to our advantage

For someone that knows how to light a scene, these details of the V1 will be quite valuable. If I need to have a car in my scene at night, slap on a bit of .6 ND gel over the headlights and they will retain their color and reduce CA while still illuminating the scene. Where highlights on a greenscreen were once a problem, now there will still be a bit of color in there for the keyer. The advantages of this kind of image are too numerous to count.

[John M. Graham]

I think those FX7 clips are pretty fugly. No offense Brian, it's the camera not you. Under well lit conditions at Fry's I like the high level of resolution and color reproduction, that looked quite fine. But, the shots in the dark TV/Theater area and the outside night shots were quite scary. The dots of grain in the store shot were huge, and I feel that the level of CA outside was pretty bad along with the lack of detail...

I'm a Sony fan, but the Canon A1 blows the FX7 away, especially in low light. Check out the night shots posted on DVi, they're very nice. I'm not sure why anybody would not shell out the extra $300 - $400 for the Canon which has XLR ports, a plethora of manual controls, instant AF, and the 30F and 24F modes.

I'm still confused as to why the FX7 is more than the FX1 at retailers, but costs less than the FX1 on the Sony Style site. IMO, the FX1 is still the superior camera, even when considering the few "advantages" and new technologies the FX7 has.

[Steve Mullen]

Originally Posted by John M. Graham

But, the shots in the dark TV/Theater area and the outside night shots were quite scary. The dots of grain in the store shot were huge, and I feel that the level of CA outside was pretty bad along with the lack of detail...

Who their right mind would take any HD camera in a dark video theater, turn-up the gain to max, and then post pix wondering what's wrong with the camera? And anyone make "evaluations" based upon such maddness is quite insane themselves.

Moreover, there is NO headlight CA at all with either of my V1's. Bright lights are perfect. And, shot correctly -- Las Vegas nitescenes are perfect -- no noise, no CA (which I have yet to see from the V1), and no burned-out colors in the bright lights.

I would disregard anything you have seen posted. Something is bogus here. You'll see no problems with nightscenes I'll post when I come back from a week at the beach. Watch camcorderinfo.com next week.

[Brian Rhodes]

I had the FX-7 for 7 hours. The gain, shutter and white balance menu is similar to the HVR-Z1U but different. Most of the night clips were shot with shutter set @ 1/60th with gain set 0-9db. I shot 20min of video with HDR-FX7 and I am for from and expert with this cam. I have other clips of office buildings gain set 0-3dB produce no dots of grain but the shoots were dark. I agree the FX1 is little better in low-light. I also took my HDR-FX1 and the HDR-FX7 to an open field with tall grass MID-DAY and shot some footage the HDR-FX7 won hands down more detail less blur. I also like the feel and the LCD size of the FX7 which is the same LCD size as my FX1. I have shot weddings using my HVR-A1U as a third camera in low-light and have had pleasing results and it has a 7 LUX rating. Normally I would not have posted footage with only a short time and not being familiar with the cams settings but I have not seen any FX-7 or V1U low-light footage on any of the forums. By the way I pre-ordered my VIU today from B&H the salesman said it would be in before Christmas.

[Thomas Smet]

Steve,

While I am not judging your judgement of the camera in any way you kind of have to see things from our point of view.

So far all we have heard from a hand full of people is how great their shots look but we have yet to see any of those shots.

The few shots we do have only look so so and the response we get is that the camera doesn't really look like that.

The FX7 is out now and for the most part it should be the same as the V1 but the shots I have seen have not impressed me very much.

Where is the footage? I myself as well as many other people will never in a million years take the word from anybody on how well a camera looks. You guys keep saying how great the camera is but we have yet to see any of this.

There are dozens of shots from the Canon A1/G1 camera that have been on DVinfo weeks before the camera even came out. The people can clearly see how good the Canon looks even though it uses "old school" CCD's.

I sit here eagerly waiting for some of this so called proof of how good the camera is and wonder why it is taking so long. If the video is for a review then somebody throw us a bone and shoot a few extra samples to show us.

[Steve Mullen]

Originally Posted by Thomas Smet

I sit here eagerly waiting for some of this so called proof of how good the camera is and wonder why it is taking so long. If the video is for a review then somebody throw us a bone and shoot a few extra samples to show us.

DSE footage has been seen by hundreds, if not thousands, of folks. It's not like anything is being hidden. The fact that internet footage is not available may, or may not, say something about what some feel about posting video on the net.

In any case, the V1 is not yet shipping and so the V1's are prototypes!

I haven't even had the time to see all my V1 video on a big screen. And, I'm waiting for the Vegas plug-in to edit 24p. I don't expect to see V1 24p on my HDTV until near the end of Dec.

Moreover, some of us publish stories as a job. When our stories are ready to be published, folks can read and see shots. There's going to be lots of coverage in December.

[Douglas Spotted Eagle]

Originally Posted by Thomas Smet

Steve,

While I am not judging your judgement of the camera in any way you kind of have to see things from our point of view.

So far all we have heard from a hand full of people is how great their shots look but we have yet to see any of those shots.

The few shots we do have only look so so and the response we get is that the camera doesn't really look like that.

then somebody throw us a bone and shoot a few extra samples to show us.

Well...some of us shoot video for a living, and in the "real world" believe it or not, it's not appropriate to post footage that belongs to the client. The footage we shot has bee turned over to (mostly) fairly high end clients.
The camera ships in December. The proof is in the putting. You're not happy with what's been uploaded (even though the same shots caused a production team from Discovery Channel to contact us about acquiring similar footage), so what's the point for those of us that have had the camera to beat our brains out trying to find a shot that pleases *you?*
Maybe the best idea of all, is to hold your horses, wait until the camera ships, and check it out for yourself.