I'm sticking with my F3

[Dennis Hingsberg]

If the F3 and F5 do indeed share similar "true native" ISO's what's quite impressive is the fact that one is four times more dense in pixels! (or approximately 4 times). That's quite astonishing.

Alister do you have any opinion that side by side the F65 produces anymore "cinematic" images than the F55? For me cinematic images means proper use of a lens, good lighting and art direction, etc.. but lately I hear some criticism that the Sony's have a particular video look as opposed to some other cameras.

I can only suggest this is a little true until of course you cross over into a world of 16-bit bit-depth and raw RGB sensor data.

[Alister Chapman]

Yes, im surprised that given the pixel counts the base sensitivity is so close. Same goes for the FS100 and FS700 which to the best of my knowledge share the same sensors as the F3 and F5, yet only appear to have about a half stop difference for similar noise levels.i suspect there is some very heavy noise reducing occurring on the new sensors.

I think the problem the F65 has is that there is so much resolution and so little noise that the images have a look that is quite different to film. It's not like video either, more like the still frames from a top end DSLR which also doesnt really look like film. The F55 isn't quite so in your face sharp as the F65 but again it's very noise free. I think part of the issue is the lack of grain and noise which can make the images a little lifeless.

Both cameras (and the F5) produce a clean blank canvas to play with. You can make what you wish with that canvas. I have to say that 4k on a large 4k monitor is just a little bit too real for me. I can see things on the monitor that I would not see in real life. But that does give you the freedom to selectively filter the image in post. The extended colour gamut of the F55 and F65 does give a very rich and deep colour palette that is deeper than most film stocks, so there is less of a "look" to the images than you would have with a film camera or camera with a narrower gamut.

[Chuck Fishbein]

I'm going to take a chance on the F5, I played with it at a demo in Boston last week and it looked and felt really good.

I have no immediate need for 4K, but I want to be experimenting with it long before client's needs arrive, so I'm not terribly concerned about the firmware schedule. It took me a long time to get comfortable with the F3, but once I was, the camera really did right by us. So, I'm assuming the learning curve will be similar.

Like Dennis, I like the idea of 50 422 onboard and 120 fps. The Nano and Samurai have served well, but it's still another thing to schelp along. It will be nice to be able to adjust speeds and not have to adjust the external recorder. Also looking forward to the new OLED EVF. Like the Nano, schlepping goes for the Zacuto or Alphatron EVFs.


The F3 is a decent camera and I'll hold on to it. Beside, I'll never get back all the money I spent on accessories. Seems like the only F3 accessory that will work with the F5 is the MTF.

[David Heath]

Originally Posted by Alister Chapman

These days it's all much less clear as very often the sensitivity setting of 0db may not actually mean that no gain is being added as much is made of sensitivity when it comes to marketing. So if a camera manufacturer can add a bit of gain and then sell the camera as a seemingly more sensitive camera, this is what they may do.

I agree with the sentiment, but not with a wording of "no gain is being added". I think "optimal amount of gain for best picture quality" may be a better description. :-) And yes, I agree that is what native ISO or "0dB" SHOULD correspond to - and I suspect that for cameras like the F5/55, it is indeed the case.

But the point is that native ISO *DOES NOT* necessarily tell you anything about real camera sensitivity for exactly the reasons Alister says - it needs to be taken together with S/N ratio at the 0dB setting.

And sorry, but in camera noise reduction does not really make a camera more sensitive - fundamentally, it trades off one set of undesirable characteristics for another. There's no denying it may give a better overall result under some circumstances, but take the camera somewhere dark, really ramp up the gain for exposure, and you'll see a ropey picture - but ropey for reasons like smear, unnatural softening etc rather than conventional noise.

Electronic noise reduction is best thought of like compression, a bit like mp3 in the audio world. Superficially, little apparent drop in quality on normal listening (but much reduced bitrate!) - but try further processing mp3 audio....... :-)

[David Heath]

Originally Posted by Dennis Hingsberg

If the F3 and F5 do indeed share similar "true native" ISO's what's quite impressive is the fact that one is four times more dense in pixels! (or approximately 4 times). That's quite astonishing.

That's looking at it the wrong way. Yes, in general bigger photosites means better sensitivity, but there is (or should be!) a huge unwritten "all else equal" caveat there.

And doing a F3/F5 comparison, all else is far from equal. If they were to be debayered with exactly the same algorithm, then the resultant F3 raster would be far coarser than that from the F5, hardly surprisingly. So whilst each resulting pixel in the F3 image may have less random variation due to noise than in the F5 case, it's going to be bigger! Hence it's reasonable to expect that will make the lower actual noise level more visible than in the F5 case?

Looking at it a different way, if we are considering a resultant 1920x1080 output, then the F3 will have to do it via a conventional deBayer. The F5 will most likely do it via direct read - as Canon have described for the C300. So although each photosite in the F3 may have an advantage due to size, each output pixel will be formed from more than one photosite - each contributing noise. That's not the case with direct read for the F5 - the red of a pixel corresponds to the output of a single red photosite.

The implication is that for all else equal, such as sensitivity is far more a function of absolute sensor size than photosite size. If you fix the sensor size, and fix the output (say 1920x1080) then varying the photosite size must mean varying the number of photosites and consequently how they are processed to give final output - it's a bit like 12 is still the answer if you multiply 3x4 or 2x6.

The "bigger photosites means better sensitivity" comes from like to like comparisons. Compare a 1/3" camera with a 2/3", both with 3 1920x1080 chips and the 2/3" will substantially outperform the 1/3", and yes, it's down to bigger photosites. But here they both must use comparable processing - that's not the case in an F3/F5 comparison, all else is not equal!

As far as the wider argument goes, then I've much sympathy with saying "what's the point of selling an F3 to get an F5"? But what about potential buyers who don't have an F3 already? Using an existing F3 is one thing, but I couldn't make a case for buying one now - not new, anyway!

[Alister Chapman]

Originally Posted by David Heath

So although each photosite in the F3 may have an advantage due to size, each output pixel will be formed from more than one photosite - each contributing noise. That's not the case with direct read for the F5 - the red of a pixel corresponds to the output of a single red photo site.

Not necessarily as any averaging that occurs may actually reduce the noise level. As the noise from any given single pixel varies from moment to moment, taking an average from several pixels may result in less noise. It's a commonly used noise reduction method.

If sensitivity is the measure of how far into the dark a sensor can see for a given noise level then noise reduction does impact the cameras useable sensitivity. On chip NR uses various methods including redundant pixels to subtract noise from the image and does this with no noticeable artefacts. I agree that many types of NR do degrade the image, but seeing as all the manufacturers do it, won't tell us how much they do it then it should be considered as part of the sensitivity equation.

[David Heath]

Originally Posted by Alister Chapman

Not necessarily as any averaging that occurs may actually reduce the noise level. As the noise from any given single pixel varies from moment to moment, taking an average from several pixels may result in less noise. It's a commonly used noise reduction method.

But deBayering isn't averaging, it's a far more complicated form of processing and I'd expect more likely a noise summation rather than reduction. In "quartet processing" (as the C300 uses) then you do indeed get averaging going on in the green channel - two green photosites are combined to form the green value for a single output pixel. But here it's a simple case of adding the two together and halving the value. If both photosites have the same signal level, then it's obvious the "real" signal output will be the same as each individual value.

The value of the approach is regarding noise (and I'm not going into the maths) and yes, that should be reduced by such a process - that's the reason for doing it.

But that's completely different from a conventional deBayer.

If sensitivity is the measure of how far into the dark a sensor can see for a given noise level then noise reduction does impact the cameras useable sensitivity. On chip NR uses various methods including redundant pixels to subtract noise from the image and does this with no noticeable artefacts.

Fair point, and certain on chip methods do give a "genuine" improvement. Maybe it would be better to word my original statement more like ".......in camera software noise reduction does not really make a camera more sensitive"?

Such methods may make a picture look better at low gain settings, but take such a camera into the dark and up the gain and it may be a different story. In extreme the image will become effectively unusable. The difference is that if NR isn't used it's the noise itself that will cause the degradation - use NR, and it's the softening, smearing etc that will be the quality limiting factor.

The best practical illustration was the "noise ghosts" of the original HPX371, caused by aggressive software noise reduction to compensate for 1/3" chips. At first sight it seemed as noise free as a comparable 1/2" camera - but the "noise ghosts" played havoc when the resulting image was processed in certain ways. I was shown footage that showed the HPX371 to be effectively unusable for green screen work.

(It's subsequently been addressed by an option of turning most of the NR off - which I'd recommend any user to do - but then it's back to the noise levels associated with 1/3". You never get anything for nothing.)

[Jeff Regan]

I don't have a dog in this fight, as we currently have a C300 PL and AF100 for large sensor camera rental and don't have immediate plans to add an F5 or F55, but after getting my hands on an F5 in my shop yesterday, I'm impressed. While it's not close to fully functional, it has great DR, super clean(not that the F3 was a slouch in that area), XAVC addresses my number one F3 complaint, the codec limitations of XDCAM EX. 10-bit and 1080/60P, along with 4:2:2 are useful and I'm not a fan of external recorders. Whether the F5 in its current state is worth the extra money over a new F3 RGB is up to each person.

My biggest problem with the F5 and other Sony cameras is the lack of a scope, and the camera not being able to output zebras. The camera I've got in my shop for evaluation right now came with an Alphatron VF, so I really don't know how one would expose the F5 reliably. The other issue I have is after getting used to the form factor of the C300, the F5 and 55 seem large and heavy, especially with battery and 4K recorder. I really like the status display--just doing simple things like changing frame rates in a C300 is a pain.

I think Sony has a couple of winners in the new F's, but am looking forward to the point where they're more complete. Having said that, I can't imagine buying a new F3 now and am surprised Sony is keeping it in production, but used, it's a strong value, potentially. I knew when we invested in a C300 it would not have a long time horizon in the rental world, but the same is true of the F3. Let's hope the F5 and 55 can provide a longer shelf life and that Sony keeps developing them above and beyond what's already scheduled.

[Alister Chapman]

The F5/F55 will get some decent metering tools. At the moment Sony are canvassing owners and potential owners on what they want. Currently most are asking for a waveform monitor, spot meter is also popular. For the moment I'm using a light meter and/or my TV Logic 5.6" LCD with waveform display. If your shooting in Cine EI Log you can use the P1 LUT and zebras.

[Jeff Regan]

Alister,

So Zebras are available via HDMI or one of the HD-SDI outputs? I also can't get focus to expand on an HD-SDI output. I think using a non-Sony EVF is a non-starter on the F5/55.

Why doesn't Sony get that waveforms are useful built into cameras? Most everybody else does.

[Alister Chapman]

I'm pretty sure waveforms will come to the F5/F55 this year. It's essential that we get either waveform or spot meter.

Zebras are not available on the camera outputs, this is true of most Sony cameras. Most 3rd party viewfinders have internally generated zebras (including the Alphatron) and you can use those. The expand focus function is done by the viewfinder itself and the scaling depends on the resolution of the finder, which is why it's not on the output.

[Jeff Regan]

Thank you Alister. I didn't go into the Error/Zone Check menu, thinking it was false color and not having enough time with the camera rig. Zebras will certainly help until there's a waveform and I knew the Alphatron did focus expansion. The TV Logic display is pretty high quality, so hopefully this should work well in lieu of the Sony VF options.

Regarding AWB, I'm surprised how cumbersome that is compared to any camera I've used, but hear that it will be put on a user button at some point.

Another issue for me is not having external audio pots. No big deal if setting a level via tone from a mixer, but what about those times when there's no sound person or mixer? Seems like an odd thing to omit and not addressable by firmware.

[David Heath]

Originally Posted by Jeff Regan

I don't have a dog in this fight, as we currently have a C300 PL and AF100 for large sensor camera rental and don't have immediate plans to add an F5 or F55, but after getting my hands on an F5 in my shop yesterday, I'm impressed.

Jeff - don't know if you've seen this, and it is referring to the situation in the UK, but if you're in the camera rental business you may find it interesting. Televisual | NEWS & COMMENTS

Most of it refers to what happened during the last year, but maybe the most interesting bit is at the end - the table of "planned camera investments for 2013" for many of the big hire companies.

As the article says (just before the actual top ten):

What’s very clear is that Sony’s two new modular cameras – the F5 and F55 – are going to be well placed in next year’s top 10 rental list. As was the case with the Canon C300 last year, it’s almost a full house for the Sony F5 and F55 when you look at the investment plans of hire firms this year. Hardly any other cameras get a look in, aside from a sprinkling of Canon C500s and Sony PMW-200s and a few more C300s and Arri Alexas.

[Jeff Regan]

David,

Yes, I own a small rental house in the SF Bay Area. We have 1/3", 1/2", 2/3", micro 4/3" and S35 cameras and various recording codecs that go with them. I do believe different markets have different tastes. The C300 seems to be universally popular around the globe, but, for instance, Panasonic is absent from the UK list, yet had been quite popular in my market(this is changing due to their lack of S35 offerings). You would be hard pressed to find any Canon small format camcorder here. Sony 500's, 700's and 800's are also thin on the ground, although popular in other US cities, primarily for broadcast/cable.

A larger, high-end rental house in my market sold three of their Alexa's and have purchased two F5's and two F55's. I'm told by them that the F55 has superior DR, lower noise and more resolution than Alexa. The other high end rental house in town has also taken delivery of an F5 and F55.

In some markets, like New York, an F3 is a harder sell to higher-end clients than an Alexa or RED. The F55 will take a bit of time to win these clients over and I'm skeptical about the day rate being high enough for a fast ROI. The F5 seems like a safer investment.

Most of my clientele are corporate, and after spending some time with the F5 a few days ago, I feel the C300 is still the better camera for my clients from a cost, size, ease of use, and workflow standpoint. I am troubled by the lack of exposure tools like waveform/vector, no audio trim pots and buried auto white balance. However, I am very impressed by the image and XAVC codec, menu handles like flare, shading, real multi-matrix and paint box capability.

Obviously, I have to give serious consideration to adding one of the new F's, but am hoping to wait for some of the firmware updates and for the cameras to build positive reputations. Time will tell, but the shelf life with any of these large sensor cameras keeps getting shorter for rental houses. Our remaining AF100 is not working much these days. It's here for the clients who want a C300, but can't afford it.