HF G50 - Video Quality

[TheG33kman]

Hello. I have a G50 with some unexpected behavior that I'm hoping to get a second opinion on.

The video quality of my G50 appears to be sub-par, specifically with what appears (?) to be sensor noise.

Videos are worth 10,000 words. Here's a sample shot:

https://drive.google.com/file/d/13Q5_W_ROG5L3bLIexwvlTZG8M4Cz8vJU/view?usp=sharing

I used manual: 1/250th, 2.6 aperature, 0 gain, 4K/24p, direct to SD card.

Google overly compresses these things and will likely mangle some of the problem out, so I added a 2x and 3x zoom after it in post to show the problem more clearly. But really, download the original file and its much more apparent - especially when viewed on a 4K display.

All over the image there is a shimmering haze of random speckles/distrortions. It's much more bothersome on clean edges where the border appears to wiggle as the pixels shift position, and on bright portions where dots come and go with a high-freqency. The closest analogue I've seen to this is when using a DSLR with high ISO in night shots. But this shot is in ideal conditions on a sunny day in shade, outside, no gain. Filming inside (my primary use) with 1/60th, 2.6, 12 gain, on 1080p through HDMI-out the picture gets worse as the speckles become larger.
FWIW, there are 4K shots from the G50 on YouTube that don't seem to be anywhere near as bad. And even when uploading mine to YouTube, it still looks worse even through the compression.

Other things I've tried:

• Filming the brightest white snow in the sunshine
• Auto setting vs manual
• Plugged-in versus on battery power
• Using it "cold" versus warmed running for an hour.
• Record direct to SD card versus through HDMI out
• Disabling stabilization

So bottom line:

Is there any setting I might be missing that could cause this?

Am I just being overly picky and expecting too much for what was "only" a $1,100 retail camcorder?
Or is this problem so abnormal that I should get it repaired?

[Inapickle]

Yes it's noise, one of the limitations of deriving 4K video images from a relatively small (1/2.3") single chip CMOS sensor. Of course you'll see it more in the shade and shadows and when you zoom in (digitally enlarge) in post. As to why it might to you appear worse that other HF-G50 footage posted YouTube, I don't know. Probably most of what footage is available on YouTube is recorded with full auto exposure anyway. 

How does it compare if you shoot with full auto exposure (Program AE) and auto AGC ? Don't know about the HF-G50, but I've seen ramped gain tests done with other small sensor 4K camcorders where the noise suppression algorithms appear to abruptly kick in above a certain gain level, presumably because they are engineered to operate optimally at relatively high levels of gain.

[TheG33kman]

Good point on the sensor. I didn't realize it was that small in this model. I just wish there was a way to objectively measure the noise level, so that I could categorically say what's normal and what isn't.

Same level of noise in auto modes. It's actually relatively good at choosing the right parameters. It just doesn't help the image quality unfortunately.

[Casey415]

Interesting post. Helps to confirm why I spent nearly double to get the G60 over the G50. Even though I have no prior experience with camcorders (I only know DSLR cameras), I did know for certain that a larger 1 inch sensor on the G60 means better image quality, especially in low light, hence lower noise level too. Best Buy dropped the price of the G50 by $100 so it's now selling for $899. I got my G60 at Best Buy when they had a $200 discount for $1499. At some point, I may need a second camcorder. I'm debating between getting another G60, or maybe the G50 if the price keeps dropping. Or maybe pay extra for the XA50 for the top handle audio ports. Or pay even more for the XF series for the larger 3.5" LCD screen.

[Inapickle]

Has anyone actually compared the HF-G50 and HF-G60 in low light ? I'd love to see a well controlled test. Yes, a 1" sensor should in theory have intrinsically better low light performance than a 1/2.3" sensor, but bear in mind that 1" sensor cameras also tend to have a shallower depth of field. I've seen some superb 4K 60p footage shot in daylight on a Canon GX10 (which I think has the same 1"sensor as the HF-G60) but after the sun went down (in the same video) I wasn't so impressed. Subjects and objects in the immediate foreground were well resolved whilst just a couple of feet back (under the same available light) much of the subject matter was fuzzy, smoothed - lacking definition. Low light 'performance' ( i.e.perceived image quality under less than optimal light) is not just about noise levels. 

[t_dhawkins]

Sent my G50 back for the same issues. That is, extremely "noisy" video when panning the camera, especially on objects with lots of lines or strong patterns.  Being a amature, I just called the phenom a "Jittery" image. Noisey is actually more to the point. It was hard to view.  Wish there was a way to get it out in the camera or fix it in post but all I could do did not work. 

I rolled the dice on a new G60. IT has the same noise in 4k but improves a bit in HD. Pretty much goes away with image stabilization turned to off, Dynamic is usable but standard is out of the question. I only have a litle video shot in my back yard for now but I plan to take it out later for a good workout. I'll decide if I like it by then

[Casey415]

Hi there. Thanks for your input. I'm slightly confused on something you wrote. With your new G60, noise level is reduced or removed when you turn image stabilization off? So when you turn on any variation of image stabilization setting while filming with the G60, there is noticeable noise in low light?

[t_dhawkins]

Nope, was referring to standard low light on a G50 vs the G60 but not under controlled curcumstances. I just took the G60 into a dark hall as I had done previously woth the g50. No comparisson. The G60 is great.

[Inapickle]

t_dhawkins

"Sent my G50 back for the same issues. That is, extremely "noisy" video when panning the camera, especially on objects with lots of lines or strong patterns.  Being a amature, I just called the phenom a "Jittery" image. Noisey is actually more to the point. It was hard to view.  Wish there was a way to get it out in the camera or fix it in post but all I could do did not work."

I thought when you sent the HF-G50 back it was because motion looked 'stuttery', which I took to mean the 'motion cadence' was not smooth. Now I'm not sure what you were referring to - was it the 'sensor' noise levels that bothered you, like the noise seen in the images that TheG33kman posted ? What you describe sounds more like aliasing, which may appear as shimmering of fine detail, stepped or jagged edges, or moiré patterns. You might see some aliasing in a 4K video on a 4K display, but it is usually more evident when downscaled to HD. Do a web search on video aliasing and you'll see many examples of what I am referring to and why it occurs. Actually when I downscaled some 4K clips, that I had recorded on a XA50 in a camera store, to HD there was surprisingly very little aliasing, which is a good thing. The 4K images coming off many cameras, and some camcorders, are just too pin-sharp and, in my opinion, unrealistic and require additional softening/filtering to avoid conspicuous aliasing when downscaling to HD. 

Do you still have sample clips from the HF-G50 that demonstrate what you were referring to ?  

"I rolled the dice on a new G60."

OK, now I'm officially jealous. All donations will be willingly accepted

[Inapickle]

t_dhawkins

"Sent my G50 back for the same issues. That is, extremely "noisy" video when panning the camera, especially on objects with lots of lines or strong patterns.  Being a amature, I just called the phenom a "Jittery" image. Noisey is actually more to the point. It was hard to view.  Wish there was a way to get it out in the camera or fix it in post but all I could do did not work."

I thought when you sent the HF-G50 back it was because motion looked 'stuttery', which I took to mean the 'motion cadence' was not smooth. Now I'm not sure what you were referring to - was it the 'sensor' noise levels that bothered you, like the noise seen in the images that TheG33kman posted ? What you describe sounds more like aliasing, which may appear as shimmering of fine detail, stepped or jagged edges, or moiré patterns. You might see some aliasing in a 4K video on a 4K display, but it is usually more evident when downscaled to HD. Do a web search on video aliasing and you'll see many examples of what I am referring to and why it occurs. Actually when I downscaled some 4K clips, that I had recorded on a XA50 in a camera store, to HD there was surprisingly very little aliasing, which is a good thing. The 4K images coming off many cameras, and some camcorders, are just too pin-sharp and, in my opinion, unrealistic and require additional softening/filtering to avoid conspicuous aliasing when downscaling to HD. 

Do you still have sample clips from the HF-G50 that demonstrate what you were referring to ?  

"I rolled the dice on a new G60."

OK, now I'm officially jealous. All donations will be willingly accepted

[t_dhawkins]

Correct, the G50 is "jittery" and I did send it back. Just took the G60 out for a spin and it is dead on. A little jitter in 4k but I guess thats over sampling, otherwise a very nice HD video. I shot no IS, Standard and Dynamic IS ....all were very acceptable.

I prefer Dynamic. I will recommend the G60. G50 just a piece of.......