Documentary Project: Camera Advice Needed Af100, FS100 or GH3/4

[KINOKS]

In which case it's either completely renew the lens set or a case of an optical adaptor being essential. Possible yes, but not as nice as lenses designed for the format.

Why not? Speedbooster brings better MTF, a stop of light and a wider FOV. Doesn't a S35 lens generally perform better on a smaller sensor when used with a speedbooster then on a s35 sized sensor without a speedbooster? Regarding IQ that is...

I plan on aquireing a 35 - 100 panny for ours GH4, bot not for the sake of IQ but because of IS and the focus trickery...

 

[harddrive]

Why not? Speedbooster brings better MTF, a stop of light and a wider FOV.

To an extent, but it can't give you something for nothing. The theoretical best it will do is give the same optical performance on the 4/3 sensor as the lens will give on the s35 sensor it's designed for.

The FOV may be obvious, but the MTF maybe less so. A given lens will give better MTF on the smaller sensor via Speedbooster in terms of lines/mm - but the smaller size of the sensor means the OVERALL resolution is (theoretically) the same as when used on the s35 sensor as intended. If numbers make it easier, let's say a lens resolves 4000 lines across a sensor width of 2cm - which would be 200 lines/mm. Now, use it on a sensor 1cm across, and an ideal optical adaptor will still give 4000 lines total - but now across 1cm and hence an MTF of 400lines/mm!

As for aperture, the smaller image will be brighter (so yes, an effective bigger f stop) - but the smaller photosites on the smaller sensor (assuming all else equal) will reduce the base sensitivity of the sensor cf s35. You're back to square one.

What it means (**theoretically**) is that with all else equal, if you took the lens off a s35 camera, didn't touch any settings and put it on a 4/3 camera via an optical adaptor, then in all key respects (FOV, dof, resolution, sensitivity) you wouldn't be able to tell the difference of which was which just by looking at the two camera outputs.

Theoretically! In the real world, then any such adaptor will - must - bring losses. That's in terms of resolution, light absorption in the glass, optical distortions and aberrations etc etc. And the more you pay, the better it will be. But it can never be perfect. Whether it's good enough depends on a lot of factors, but the principle is why I said "not as nice as lenses designed for the format."

Doesn't a S35 lens generally perform better on a smaller sensor when used with a speedbooster then on a s35 sized sensor without a speedbooster? Regarding IQ that is...

Simple answer is NO. For the reasons above. It WILL give higher MTF at the sensor (in lines/mm), and it WILL be a higher f-stop (brighter image), which is what confuses many people. But as above, those factors are cancelled out by the bigger sensor needing a lower lines/mm figure to give the same basic resolution, and the bigger sensor having bigger photosites and a correspondingly better sensitivity. ( So same low light performance, theoretically.)

And that assumes ideal convertor performance.

Practically, the performance on the smaller sensor can only be worse on the smaller sensor with Speedbooster, than used without on the format size the lens is designed for.

 

[David W. Jones]

They may work - but unless you use an optical adaptor there will be a magnification effect, as I'm sure you're aware.

An 18mm lens designed to cover a S35mm size sensor will give the same field of view as an 18mm lens designed to cover a M4/3 size sensor when used on a M4/3 camera.

But at the end of the day, your M4/3 lenses will not work with a S35 sensor camera in the event you decide to go in that direction in the future.

I only own 1 M4/3 native lens, which was the 14-140 that originally came with my GH1.

All of my other lenses will cover S35, and the majority will cover full frame. That way I'm not tied down to a single format, with the need to sell and purchase new lenses with each camera change.

 

[harddrive]

An 18mm lens designed to cover a S35mm size sensor will give the same field of view as an 18mm lens designed to cover a M4/3 size sensor when used on a M4/3 camera.

No - not true. Angle of view is determined by focal length and sensor dimensions - see http://en.wikipedia.org/wiki/Angle_of_view if you want all the maths.

In simple terms, scroll down to "Sensor size effects ("crop factor")" and it says :

As noted above, a camera's angle of view depends not only on the lens, but also on the sensor used. Digital sensors are usually smaller than 35mm film, causing the lens to usually behave as a longer focal length lens would behave, and have a narrower angle of view than with 35mm film, by a constant factor for each sensor (called the crop factor).

Keep focal length the same and make the sensor size smaller and the angle of view must be narrower - unless you use an optical adaptor.

The other issue will be with resolution. If the s35 lens is engineered to give (say) 4000 lines of resolution across the horizontal with an s35 sensor, if you halve the linear sensor size it will only give 2000 lines. This is unlike the optical convertor case above where the whole resolution is theoretically preserved.

As regards aperture, nothing will change - but you've now got a much heavier, bigger and (probably) more expensive lens for a given focal length and f stop than if you'd gone for one optimised for the smaller sensor size in the first place.

 

[David W. Jones]

I don't think you understand what I said.

Mount an 18mm lens designed to cover a S35 size sensor on a M4/3 camera

Now mount an 18mm lens designed to cover a M4/3 size sensor on a M4/3 camera

Your field of view will be exactly the same!

 

[KINOKS]

Mount an 18mm lens designed to cover a S35 size sensor on a M4/3 camera

Now mount an 18mm lens designed to cover a M4/3 size sensor on a M4/3 camera

Your field of view will be exactly the same!

Khm... No, if it's written 18mm on the lens then it should be 18mm period. An 18mm lens on a m43 format will have a narrower FOV then on a s35 format. Since when do manufacturers take into account the formats crop factor when writting down the lenses focal lenghts on the lens body??

 

[KINOKS]

Practically, the performance on the smaller sensor can only be worse on the smaller sensor with Speedbooster, than used without on the format size the lens is designed for.

Thx for the great explaniton

It seems like the only thing you get when using bigger lenses on a smaller format with SB is a stop of light. That is if the MTF of the native m43 lens is that much better then the bigger lens. Right? When comparing using m43 glass or s35 glass on a m43 sensor. Not comparing s35 sensor with m43 sensor.

 

[FrameFarmMedia]

Thx for the great explaniton

It seems like the only thing you get when using bigger lenses on a smaller format with SB is a stop of light.

With the speedbooster you gain a stop of light and .71x wider FOV. Two very important things for a smaller sensor.

 

[yzw200]

Hey All,
I truly appreciate all the information and assistance you have provided. I've decided to go down the AF rabbit hole and has an additional followup question: While I know the AF100a was not much of a seller as the upgrade was viewed as not worth the cost--as someone purchasing now, would you go for the 100a with a Ninja Star to get 4:2:2 10bit at about a $500 difference from or an AF100 (with software update)+Ninja.

Thanks again for all your help!

Yisrael

 

[KINOKS]

It's not real 10 bits. It's 8 bit with dithering. And as for the Ninja; must be the worst recorder I ever used. Bad screen and functionality, hanging up all the time, slow, etc.

 

[John B.]

I've had really good success with a Samurai Blade. Atomos just lowered the price on them, too. With the Samurai, you can go with a fully professional connection: HD-SDI vs the less field-worthy HDMI connection on the Ninja. If you still prefer the Ninja, get the Blade version, as that incorporates the much improved screen.

 

[yzw200]

Kinoks,Sorry for my ignorance but I'm coming from a dvx background...can you explain the practical difference between 10 bit and 10 bit dithering? I understand 10 bit is a major upgrade from 8 bit, are you saying the practical difference between an a 100 and 100a is negligible? Thanks for breaking this down for me. Appreciate it.

 

[harddrive]

I understand 10 bit is a major upgrade from 8 bit, are you saying the practical difference between an a 100 and 100a is negligible?

10 bit *MAY* be an upgrade from 8 bit, but it's not as simple as it may seem.

What 10 bit *can* bring to the table is the ability to define video levels more precisely - which can be very valuable in post work etc in particular. **BUT** it's only effective if the precision is there in the first place, in the original video signal. And unless the camera is very noise free, that won't be the case. In many cases (and I'll include the AF100) the video noise level is such that encoding to 8 or 10 bits won't really make a lot of difference.

Move onto a far more expensive camera, and especially if you're using something like s-log, and it's a very different story. That's where 10 bit certainly will offer a big advantage over 8 bit encoding.

Kinok's point is that the AF100 is only really capable (in common with many other, some much more expensive, cameras) of giving a signal which does justice to 8 bit. The AF100A may have 10 bits/sample on it's output, but the least significant aren't really carrying useful information. Practically, for a given bitrate, you'd probably be better off encoding to 8 bit with lower compression.

 

[KINOKS]

Kinok's point is that the AF100 is only really capable (in common with many other, some much more expensive, cameras) of giving a signal which does justice to 8 bit. The AF100A may have 10 bits/sample on it's output, but the least significant aren't really carrying useful information. Practically, for a given bitrate, you'd probably be better off encoding to 8 bit with lower compression.

Actually no, this is not the case. With AF100a the HD-SDI output is 8 bit with two extra bits that are only dithered. As Barry said:

Correct. It's a 10-bit output, sourced from the internal 8-bit and processed ("enhanced") to 10 bits.

So there is no difference if shooting 8bit then up converting to 10 bit in post and adding some fine noise. AF100a is a 8bit camera.

 

[harddrive]

Actually no, this is not the case. With AF100a the HD-SDI output is 8 bit with two extra bits that are only dithered. As Barry said:

So there is no difference if shooting 8bit then up converting to 10 bit in post and adding some fine noise. AF100a is a 8bit camera.

KINOKS, we're basically all saying the same thing, what you and Barry say is more or less the same as what I meant by "The AF100A may have 10 bits/sample on it's output, but the least significant aren't really carrying useful information."

To yzw200, think of it in terms of a beer glass - you have the choice of a pint or a litre glass - which do you go for? At first sight it seems obvious - the litre glass will give you a lot more! But what if you get a litre glass, but only with a pint of beer in it? Same principle here, you're getting a 10 bit "glass", but only 8 bits of useful video in it!

And if you accept you're only getting a pint in either case, there may be advantages in just getting it in a pint glass. In the case of the beer, the glass is easier to hold ( ), in the case of video, the 8bit "glass" needs a lower datarate to compress to the same standard.

 

[Marino215]

So basically it's still just an 8 bit signal in a 10 bit wrapper. I've done this a lot with my pix 240/fs700 combo and there's really no discernible difference in quality. The biggest difference I've noticed (and it's not a lot) was going from 420 to 422 with a higher bit rate/different codec. Maybe It's just me but from my experience I'd rather deal with 10/12/14 bit 1080p image than an 8 bit 4k one. Once you truly experience the flexibility of higher bit rates and color spacing an 8 bit 4k image seems limiting.

 

[Adams]

Congrats on your decision Yisreal! I shot a long-form verite-ish documentary on a DVX100 (and Canon XL2) and then bought the AF100 right when it came out when I started my next project.

I've shot both (mostly) myself, as a one-person crew. I wanted to share my thoughts a budget-concious filmmaker following a similar path. I love the AF100, and you can great stuff with it. The ease of audio work with a traditional on-cam shotgun + lav, is so important and underrated in those situations. A 2nd lav attached to a zoom in your backpack can make a big difference. Keeping things in focus is definitely harder, which means it's harder to have eye contact with subjects. I mostly use a Nikon 17 - 55 2.8, as it gives me reasonable range and good light. But there is no IS, and 55 is pretty short. If I get in a car or some other bumpy place, or if I need to follow a subject from a distance, (like outside) I go with a Panasonic 14 - 140, which has IS and the approximate range of the DVX, but it's not a constant aperture - you'll often see it darken or lighten while zooming. Good luck!!

 

[den59]

To add on "not as good as the lens designed for the format", sensor stack thickness is another variable in adopting third party lenses. The presense of the glass of different thickness infront of digital sensors may/will affect results depending on numerous factors. There are good posts online on that topic, dont have links at hand, though.