HMC-150 / HMC-151E noise problems - read this

MrFluffy

MrFluffy

Well-known member
After a long and protracted problem regarding noise with my HMC-151 and the terrible service that Panasonic provided I discovered that one of the default settings in the HMC can cause you to get very noisy looking images.

The original thread is here:
http://www.dvxuser.com/V6/showthread.php?t=159062

By default, when you turn on the zerbras to get your exposure correct, you will under expose your images. The default setting for the zerbras is 80% which is too low and cause underexposer.

Go ito the menu and set this to 95% and you will get crisp clean images.

I hope this helps someone avoid the problems I had.

:beer:
 
makes sense but the 150 has 2 zebra settings that are selectable in the menu. usually you would have one set to 70-80 for your skin tone facial highlights and the second set to 100 for over exposure etc. I believe mine came set up this way out of the box. Zebra 1 set to 80 and zebra 2 set to 100.

I'm glad you were able to solve your problem and it wasnt a fault of your camera.
 
Underexposure leads to grain, yes. I usually use zebra 1 set to 70 for face highlights, and zebra 2 set to 100 or 105 to avoid clipping. On the 170 and 150 you can also pop up the waveform monitor and get a glance as to what your exposure is. If you're not using the full range, if your image is mainly clustered down towards the bottom, yes you can expect noisy images (on any camera). The upper range is clean as can be on a 170, 150, or 200A.
 
Barry_Green said:
On the 170 and 150 you can also pop up the waveform monitor and get a glance as to what your exposure is. If you're not using the full range, if your image is mainly clustered down towards the bottom, yes you can expect noisy images (on any camera).
Click to expand...
There is a waveform viewer on the 150? Where?
 
The scopes are the coolest feature. I took Barry's zebra advice years ago and hasn't steered me wrong. The 105 percent is a great setting...
 
I've always used 100% but am curious to how you adjust the zebras when set @ 105. Do you still back them off until they disappear? Or do you drop a bit lower after they are no longer on lcd?
 
Can I give a slightly different perspective re. using and setting Zebra levels. I am a stills photographer who is a long-term film, then video director and cameraman.

I feel that the OP does not understand how both zebra setting are used to choose an optimal exposure. Please excuse the length of the post. If you use the 80% bars as described, the scene likely will be underexposed, and noise the probable result, as others have noted.

First, the stills perspective. All incident light meters measure light coming to the camera from a scene. The folk at Kodak (Rocherster NY) many years ago assumed that the subject of interest in a scene reflected a particular percentage of light back to the camera—and chose 18%, judged to be the "average" reflectance of Caucasion skin. All stills cameras, more or less, adhere to this convention. If you point your camera at an 18% grey card, and adjust your exposure for that (so the metered light is in the middle of the scale) then the scene will be correctly exposed—IF the subject of interest has the same reflectivity.

But what if you are shooting an outdoor scene in snow? If you allow the meter to adjust the light to be 18% reflectance, then your snow will look grey; you have, in fact, underexposed for that scene. If you use a spot meter (typically 1 degree angle of view), and point it at (say) the cheek of a subject in your scene, then set the camera exposure to suit, then the snow will look white, as you probably want.

Tangential, but relevant, is dynamic range: each imaging device has a range from dark to light that it can reproduce. 8 stops is good in the stills world; a Fuji S5 Pro can reproduce ~9–10 stops with electronic wizardry (the sensor has both small and large pixels). So, to echo one of the posters above, IF you underexpose, relatively fewer pixels are being used to reproduce this DR; increased noise is the result.

Ideally, you want all pixels recording a value; if underexposed, there will be plenty of highlight detail, but less shadow detail (they "crush" to black); if overexposed, then the highlights blow out (reach full saturation, photons being measured by pixels-wise), so they display pure white and no detail, and the shadows will have plenty of information.

From the above, there is no "correct" exposure: only where (between black black and white white) you want to locate the DR your device *can* capture. This is why Hollywood typically lights outdoor scenes: to bring shadow detail into the DR that can be captured. Indoors, no problem: it is easy to get the 8 stops difference the sensor can capture. This is only a consideration when the DR of the scene exceeds the DR of the capturing medium.

Now to the video world. The HMC150 has two "zebra" bars. These measure brightness (luminance) values within the scene been shot. The factory settings (from memory ) are 80 and 100%. When the 80% ones are tripped (that is, on a face, for example, the cheeks have a small area/dusting of zebra bars, that is telling you that the brightest part of the face is reflecting at ~80% brightness (as assessed between between 0%, black and the darkest part of the scene, and 100%, or white, the brightest). So, this will be close to the "ideal" exposure spoken of above, but keep in mind that value (18% reflectance) assumed that the skin was Caucasion. And when you check the 100% bars, AND you have a white object in the main light, the object should trip these bars; accordingly that object will reproduce as white on playback. If this describes your scene, you are good to go.

But what if your subjects has Indian, or African-American skin? If you go by the above and adjust the iris so that the 80% zebras are tripped on a cheek, that person's skin may be *too* bright in the final range of values in the scene to look realistically dark, as you'd probably want. By using the 80% bars this way, you are telling the camera that you want that part of the scene to be reproduced at 80% brightness. Dark skin, but with detail, is probably what you want. In this case, you will need to apply exposure compensation. In the stills world, this is simple: we have those buttons dedicated to this purpose, but in videoland, you will need to check either the waveform monitor as BG suggested, or output the camera to a calibrated monitor, and adjust the iris to give you the look (in this case, brightness) you want. If you have dark and light skin (or any other light or bright subjects that you want to render realistically) you need to decide how to reproduce that range of lightnesses and brightnesses in the limited DY you have.

Alternatively, still assuming a scene of light and dark skin that you want to reproduce realistically, you could use a spot meter (I use an old Pentax one); measure the reflectance of a dark cheek, and set the lower zebra bars to that (probably ~55–60%); and set the top one to suit the Caucasian skin (the 70–80% spoken about before). Reading both will tell you if you need to reduce the light on the lighter skin, or increase the light on the darker skin—because you can only record a certain range (the DR we were talking about). Then use the WFM to make sure that you have captured the widest set of values you can, and to make sure that the brightest objects in the scene are not blowing out.

hth, kl
 
Last edited:
Wow Kit, thats an excellent explination. Thanks.

The users guide doesnt list any of this stuff. I so need a good book on the 150/151
 
MrFluffy said:
Wow Kit, thats an excellent explination. Thanks.

The users guide doesnt list any of this stuff. I so need a good book on the 150/151
Click to expand...

What Kit is describing is not so much an HMC thing as it is a filmmaking thing. 18% and dynamic range are two concepts taught in film school that takes some people a bit to understand. It's hard to care, in your first class using a film camera, what it means when your light meter gives you an f-stop of 4. Most people know that it's telling you to set your aperature to 4, but they don't understand it means to set your aperature to 4 if you want that part of the scene to be exposed to 18% grey. This is the same concept used for film-noir (among other genres). The lights will be set up in such a way that half of the face is properly exposed, and the other half is 2-3 stops underexposed, creating a specific mood.
 
I think persons complaining about noise are likely using the lower range zebras and adjusting the image either to hit the zebra, or be just lower than the point of showing the zebra.

For me, the camera performs best in the highlights, that being said, you can use the zebras above 100 and set the exposure to the point of just tripping them, you will get more info in the shadows, and this camera still has detail at 100-105 (although if it were really 100, it shouldn't, for me in practice it does) that can sometimes still be extracted in post. The latitude in the highlights is amazing, so keep it there. You really don't need special equipment if you keep it there.
If you clip, you can always add color to the highlight in post to make it not look blown out.

You can check your shadows with a spot meter. If you don't have enough light in the shadows, you can add a fill, if possible and you will be good.
 
Last edited:
Nice explanation, Kit_L, but I think you're wrong about the 18% reflectance and its relation to caucasian skin. The average reflectance of caucasian skin is not 18% - its higher. As far as I know the 18% reflectance is related to the zone system - the system that every cinematographer should be familiar with. Its aim is to allow for thinking about various elements of the frame in terms of exposure and setting the exposure (and film development) properly. The dynamic range has been split into 11 zones (0 to 10). Zone 10 corresponds to pure white, zone 0 represents black. Each zone differs from each other by one f-stop. This is where 18% is derived from - if zone 10 (pure white) is 100%, the lower zones will have the following reflectances (obtained by dividing by 1.4142 equivalent to stopping down by one f-stop):
zone 10 - 100%
zone 9 - 70.71%
zone 8 - 50%
zone 7 - 35.35%
zone 6 - 25%
zone 5 - 17.68% (rounded to 18% for convenience)
So, the 18% is the reflectance of the 5 (middle) zone (middle gray for b/w photography).

The caucasian skin reflectance is closer to 25%. As a result, in a properly exposed photograph the caucasian skin will fall into zone 6. Of course it is up to the cinematographer to decide what 'properly exposed' really means - you might want the caucasian skin to fall in zone 2 for backlight silhouette effect with some skin detail. The zone system gives you an easy hint on setting your f-stop for such effect (you will have to underexpose by 3 f-stops).

I once was told that 18% reflectance is the average reflectance of the world. Strange statement to me. If I lived in an average world I wouldn't like to be a cinematographer, I guess ;)

Even though the zone system is of great use for videographers we should rely more on the vectroscope (ASA rating becomes fuzzy term for the video camera and setting the exposure with incident light meter is tricky - the meter is still great for setting light ratios, though).

Personally, I don't like using the 70-80% zebras for setting exposure. How they relate to the way the footage looks depends too much on the gamma curve. I prefer to use the 100-105% zebra which indicates the limit of my dynamic range. This way I can decide which areas of the frame will be clipped. If I don't need any detail on f.ex. bright window I clip - no problem to me at all. But I want don't want to loose the information of the essential elements of the frame. If I have this information I can always adjust the footage in post. I only use 70-80% zebra as a guideline for checking if I pump enough light to the scene so that the faces are properly lit. But I prefer relying on 105% zebra for setting the exposure... Just my 2 cents...
 
matt s. said:
I've always used 100% but am curious to how you adjust the zebras when set @ 105. Do you still back them off until they disappear? Or do you drop a bit lower after they are no longer on lcd?
Click to expand...

Just set it in the camera options... to increase to 105. and you are totally right. I just make sure that there's nothing at 105... just barely below it and then I know I'm right at the treshold. Works really well... 100 percent isn't bad either, but I just like how precise this is...
 
Kit_L said:
...Alternatively, still assuming a scene of light and dark skin that you want to reproduce realistically...

hth, kl
Click to expand...

Hi Kit L,

Have you experimented with the DRS on the HMC150? It expands the limited dynamic range of video. Level 1 and 2 work quite well. I would like to know your thoughts on the DRS feature. Thanks.
 
Mark Von Lanken said:
Hi Kit L,

Have you experimented with the DRS on the HMC150? It expands the limited dynamic range of video. Level 1 and 2 work quite well. I would like to know your thoughts on the DRS feature. Thanks.
Click to expand...

Unfortunately not available in 24p/25p on my HMC151 :(
 
Justyn said:
Just set it in the camera options... to increase to 105. and you are totally right. I just make sure that there's nothing at 105... just barely below it and then I know I'm right at the treshold. Works really well... 100 percent isn't bad either, but I just like how precise this is...
Click to expand...

I think it is sometimes good to clip to save the precious bits for what's really important in the frame. F.ex. the specular highlights on a sunny day are very bright and trying to fit them in the camera dynamic range would require stopping down several stops... The hot windows on an interior shot (which are meant to stay hot) are another good example...
 
You must log in or register to reply here.
Back
Top