joe12south
Veteran
Definitely does. The farther you get from the "native" ISO on any camera, the DR decreases as the noise floor increases. Impossible to say until we can test, but good DR up to 1600 ISO looks likely.
GH4 GH5: Seems they haven´t thought about DR at all. Awful DR?
- Thread starter daktulus
- Start date
filmguy123
Veteran
This is one of the selling points of more expensive cameras. Arri advertises heavily the fact that their Alexa cameras are 14 stops through the entire ISO range - that's a big deal. Generally when you hear advertised DR numbers they are in best case, native ISO settings.
Point being if the GH5 actually holds 12 stops through ISO range well, that's great! And better than perhaps an A7s which ranges from 13.5-12. DXOmark rates the RAW sensor output from GH4 at only .5 below DR of A7S. There's also more to the story and a number of factors that play into the final resulting image quality and usable latitude.
dustylense
Veteran
Some of you DR freaks. Stay far away from transparency film. You would literally not have a good day.
Emanuel
Cross-Examiner
Right. People are DR addicted, that's the whole thing. Pixel peepers, that's it.
John Godden
Well-known member
That video has SUPER low contrast (DR) on all ISO settings. See my comment on youtube.
I will say that the noise looks much improved over the GH4 at the higher ISO settings. :thumbsup:
Matt Gottshalk
Resident Preditor
qftmft
joe12south
Veteran
This isn't like the resolution arms race. Having room to save a shot from over/under exposure in post is something that stills shooters have been enjoying with RAW for some time. It's totally reasonable to want a motion camera to offer this same flexibility.
Emanuel
Cross-Examiner
Pixel peeping is not only addressed to resolution, actually a pixel serves for much else...
https://www.dpreview.com/forums/post/51470841
Emanuel
Cross-Examiner
LOL
Many sensors are designed to provide their best dynamic range at their native ISO, which often is NOT their lowest ISO rating.
It appears that the GH5 ISO noise is mainly luma channel noise with little effect on the chroma signal, which means the color will be retained even at higher ISO settings.
While we don't have RAW, the 10-bit 4:2:2 capture should provide more room in post to get the image you are after. One comment I keep reading about with the Luke Neumann GH5 V-log files is how well they grade compared to GH4 footage.
The limiting factor with GH4 DR is noise. When taking photos at iso100 there is less noise and I can dig more clean shades from dark tones. GH5 has in general less noise than GH4 and using iso100 with iDynamic may give cleaner black tones. Also the new HDR profile with iso100 may give more dynamic range. It is a flat profile using the whole recording space for gradations. GH5 sensor has about 13 stops DR. With proper processing it should be possible to get that for video.
joe12south
Veteran
It's not just chroma noise that pollutes color at high ISO's. The actual chroma levels begin to drop and shift. Compression exacerbates the problem. You start to see this on the GH4 at ISO 1600 and then it drops off the cliff above that.
joe12south
Veteran
- 100 will probably not be the cleanest ISO...it's an extended range option. My prediction is that 400 will be the "native" ISO (highest DR with the least processing.)
- From everything revealed so far, I don't think we should expect the GH5 to have more than 12 actual stops of raw DR. I think the difference is that it is all going to be usable, vs the GH4 in which only about 10 was actually usable. The better processing should allow us to reach deeper into shadows and use higher ISO's, but I doubt that means a "true" extra stop of DR. If I end-up wrong about this and there is more, I'll be pleasantly surprised.
It's difficult if not impossible to say, given that he isn't shooting a DR chart.
...But if he was shooting an interview at that table, how high could the ISO go and still end up with a decent looking result ?
I'm guessing that 3200 ISO with a bit of NR would still be acceptable to most of us.