F3 through its paces

Posted on May 7th, 2012

I needed to share some feedback on the F3 now that I’ve owned one for 8 months and worked on a variety of projects with it from interviews, feature length films, artistic music videos, and even photography (yes with the F3). For a Super 35mm motion picture camera you might expect rigid processes and finagling but the F3 is far from it once you learn how to use all the features of the camera and combine them.

GAMMA MODES

I’m now only exclusively shoot using gamma modes 1,3,4 with custom white balance temperatures dialed in. This let’s me cool down or warm up a shot depending on what I need, and especially outdoors where clouds are passing by creating shade and sunlight sporadically. I seem to use 4400, 5200, and 3200 the most for some reason. Gamma modes 1,3,4 will record signals above 100 IRE making them illegal for broadcast but adjusting ‘levels’ or ‘curves’ in post, or by using a shoulder plugin or some other legal NTSC plugin in your favourite NLE you can make your signal legal again. I like to spread my signal over a max area for capture (0-110 IRE) so I know I am clipping less and given some freedom later. To see the curves I put this handy document together; www.starcentral.ca/Sony PMW.pdf

ND and GAIN

I own a full set of 4×4 and even circular ND filters but rarely need them with the F3 since between the built in ND filters (3 and 6 stops equiv.) and gain up/down settings (and the fact F3 is virtually noise free) I can accomodate a variety of exposure levels indoors and out to get my desired f-stop and resulting DOF. Never in a million years would anyone ever want to use GAIN to shoot (at least I never did in fear of introducing noise) but since the sensor of the F3 is so clean at high ISO, I can switch to the MAX built in ND filter then use my GAIN to bring the image back to where I want it, for example if I want to keep my lens iris at a specific f-stop. Indoors I’m typically shooting with f5.6-8 or 11 on wide shots and f2.8 on medium or close up shots. Outdoors are similar, but I lighten up on the DOF and tend to shoot closed down by 2-3 stops. (backgrounds are usually much further back anyway or infinite, whereas indoors you are in closer quarters).

SHOOTING THE SUN

The first thing I started shooting with the F3 were sunset/sunrise time lapses. A very good option I found for shooting the sun in the sky under custom picture profile is to select the FL setting under MATRIX. Most other MATRIX settings create a fringing blaring ring of ugly yellow chroma artifacting. Thankfully the FL mode seems to solve this.

EXPOSURE

Everyone’s process for image exposure is different, but for anyone who has read my post here on using a light meter for digital image acquisition knows I am a big fan of using it as my guide. I won’t get into too much detail how/why to use one here but will say its especially useful for scouting locations before even bringing or setting up the camera. Its just a guide, it works for me, and I will never stop using a lightmeter on my shoots. With my lenses I’ve calibrated my F3 @ 24fps 1/48th or 180 degrees 0db, 0 ND to be 800 ISO, or require 42 footcandles of light @ f4 to expose a grey card at 50%. I can’t and don’t drag around ext. monitors with waveform monitors or other tools built in so I am using what I have.

After I take a meter reading I set my lens accordingly, then use ZEBRA to see what’s over 100. Blaring hot sources or blown windows a little over 100 IRE do not bother me most of the time. Its what works for the scene (ie. do you really want to see what’s outside that window for your indoor shot? It can be distracting if not out of focus.) I also hover the built in marker function over the subjects face to see where it falls, I like 40-60 IRE depending on skin color and lighting angle. Last I cycle through my gamma modes to find one that evens out the balance of the scene between subject, background, and other elements.

Gamma 1 is great for high contrast situations. I would use this for a city shot with clouds in the shot, or maybe a subject standing behind a brigher background.

Gamma 3 is like REC 709 but illegal. Its great to record over 100 IRE and is a middle of the road curve I try to shoot 85% of the time with.

Gamma 4 is really good for low light situations, like indoors with limited lighting, or even with lighting. Interestingly, I’ve used it outdoors to lift the levels of the skin without increasing the brightness of the surroundings. I like this one a lot.

Why gamma curves? If you do it in-camera you don’t have to do it in post. And since I record to SXS cards (which is 8-bit) any grading I do in post I want to use to stylize my images, not fix my exposure. This is a BIG THING people when people talk about 8-bit not holding up for post. If you nail your exposure what you have left in post is some room to do some grading and styling work without tearing apart the image. My last check is the histogram but I don’t take what I see there as gospel. If you are shooting a high-key scene then YES your waves will be bunched up to the right for example, so one must be careful using a histogram to judge exposure.

Below, a frame taken from a music video I directed and shot recently for artist  Mike D., song titled BACK IN TIME. Metered f8 (shade side) on subject, Gamma 3 used, 5200 white balance, built-in ND1 filter .9 (3 stops), taking lens set to f2.8.

Posted in Filmmaking

Lightmeter use with the F3

Posted on January 4th, 2012

I have been using my Sekonic Zoom Master L-508 lightmeter in conjunction with my F3 since I got the camera last year. One of the things I really wanted to be able to nail down was getting the best exposure and consistent exposure for shoots where I may be limited in external tools, or need to shoot a light rig, or as it often happens there is just not enough time between setups and you want a quick “gauge” of what’s going on with your light. I’ve learned to be able to rely on this, trust it, and have it work for me.

I’m not going to suggest to anyone that you ONLY use a lightmeter in place of using your eye and some of the tools available on the F3, but for example using zebras to gauge exposure on skin can be extremely inaccurate because of the way light could be hitting the face. Are zebras just starting to appear on the bridge of the nose okay, or cheek, forehead… ? Just that difference alone can amount to one or more stops in adjustments and once you start moving your setup or subjects around you can really start to get inconsistencies I find.

If you watch the F3 exposure video on Abelcine’s website you can see that natively the F3 has an ISO of 800 at f4 to expose middle grey with a setting of 24fps, 1/48th shutter (or 180 degrees), 0db gain, and no ND filters engaged. As I been shooting in a variety of conditions under direct sun, overcast, indoors with controlled lighting, and outdoor snow conditions I have found that I can set my lightmeter to 24fps 180 degree shutter, take an f-stop reading and 9 times out of 10 it is accurate to my lens iris. As even with film, one has to obviously take into consideration what’s happening in the rest of the scene in terms or brightness or darkness and compensate accordingly.

So for example in this scene I walked through the trail to find the brightest patch of light coming through the trees. I measured f8 on my light meter and then set the built-in ND filter to “1″ which is 1/8 ND (0.9) – reducing 3 stops from f8 to f2.8. Because there was lots of snow on the ground and in the trees I stopped down the iris an extra stop and set it to f4. Below is a frame grab from the scene,

A quick edit of clips from the unfinished promo video can be seen here,

During one take I did actually open the iris to f2.8 and it looked fine in the LCD but watching it later on a large screen the scene was semi-blown and the shot looked bad to me, so you have to use judgement with a lightmeter like with any tool. Had I only used zebras to show me when the snow started hitting 100 it would be debatable how much or little of the snow should show up in stripes, and worse at the expense of changing the exposure on my main scene elements (the actor and horse)!

Using a lightmeter is really no different than using a grey card so if you learn to use your meter with your camera you should easily know where 50 IRE is. From there you can compensate your iris accordingly depending on the color of skin you are shooting, look/style you are after, and lastly compensate for your overall scene if needed (primarily only for outdoors). One thing to be careful about as others have mentioned, is that when you start using the gamma curves you effectively start moving your middle grey point around which will skew things around. I find I can still pretty well match my iris f-stop to my lightmeter, then go through cinegamma 1,3 & 4 until I get the picture that looks the “best”. I avoid cinegamma 2 because it clips at 100 but if you need an immediate legal signal then cinegamma 2 is good.

Some examples using lightmeter:

f2.8 metered and 2.8 on lens:

f22 metered ambient*. 1 stop loss from overhead diffuser panel, 4 stops loss in ND filters, f4 on 85mm lens.

*This was metered away from the sun because subjects were more backlit. Typicially this works out to about 2 stops less then if metered against direct sun.

f2.8 metered, lens set to f2. -1 stop underexposed purposely:

Posted in Filmmaking | Tagged as: , , , | Leave a comment

So you think you need S-LOG?

Posted on November 7th, 2011

So Long Music Video shoot with Sony F3

This is for all the non S-LOG people out there who might believe it’s needed to capture and grade images to get beautiful results. I know first hand what it is like to want the best of the best all the time, and even then still feel behind with your equipment in one way or another.

There is bickering and arguing over the need for S-LOG and recording it off-board to capture 10-bit 444 RGB output versus recording it internally to SxS 8-bit and getting the extra “look” and lattitude of S-LOG, or just forgetting S-LOG all together.

People have sworn that you can’t grade 8-bit but not sure I agree. It depends on the quality of images you start with and the level of grading that is intended. I just finished shooting at a beach this weekend in direct sun with no clouds and let me know what you think of these images – all were recorded non S-LOG directly to SxS cards in 35Mbps 4:2:0 XDCAM format and “graded” from the originals.

So Long Music Video shoot with Sony F3

S-LOG is said to give much better skin tones and color rendition that can’t be described but since I don’t own S-LOG nor have I shot with it yet, I simply can’t attest to that. I have seen some bad examples of S-LOG too. Its all very subjective. At this point I only have the specs to go off of and then ask myself if I really need 14.5 – 15 stops of dynamic range and on top of it spend $3k + $6k to record it off-board? Will my clients notice? Will my work really benefit from it? Would a beach shoot like this had been a disaster with all the extra off-board gear needed for S-LOG?

So Long Music Video shoot with Sony F3

In the end I think with proper lighting and good base exposure you can really squeeze a lot out of the F3 and no one should disagree with that since people have shot beautiful images using cameras with less dynamic range capability.

So Long Music Video shoot with Sony F3

Posted in Filmmaking | Tagged as: , , , , , , | Leave a comment

Footcandles to expose image on F3

Posted on October 20th, 2011

Since I left behind the world of shooting with 35mm adapters where I was working with effective ISO’s from anywhere between 50 and 160 to a bright new world of 800 ISO with the F3, lately I’ve been obsessing over different light fixtures and their output performance combined with what’s exactly needed to expose a “good” image with the F3.

I did some reading of course and really liked the way ARRI was rating the “performance” of their lights. Most manufacturers rate their light output in lux or footcandles at set distances, but since light output is inversely proportional to the square of the distance from the point source ARRI additionally provides a “performance” number for spot, medium and flood and this let’s you do your own math based on distance to calculate out your footcandle power.

I like this, because I can make a quick reference guide comparing “performance” on everything from Kino Diva Lites, Lowel lights, ARRI Tungsten, ARRI HMI, Light Panels… well you get it – practically anything!

The formula is listed on every photometric data page on ARRI’s site. Basically, footcandles = performance / distance X distance. So for example, an ARRI 200w HMI has a medium focus performance number of 20,000 (there are separate performance numbers for spot and flood). To figure out how many footcandles I’d have at 20 feet just divide 20,000 by 20ft x 20ft (or 20ft squared basically). You get 50 footcandles. While this is an exciting way to compare lights and also figure out the approximate illumination you will have at set distances, what I really wanted to know now is how it would translate into shooting with the F3 and its crazy noiseless ISO capabilities?

My tests revealed that I would need only 42 footcandles to expose middle grey to 50 IRE @ f4.0 on the F3 using the following camera settings: 0db/800ISO, 1/48th shutter (or 180 degree), 709 Rec mode, no ND filters engaged. To do the test I used a 42″x32″ softbox to illuminate a 18% grey card and filled the complete frame using a 85mm lens set to f4.0 on the F3. I moved the light further/closer away from the grey card until the marker function on the F3 read exactly 50%. I then took a meter read using a Sekonic 508 set to footcandle mode and measured out 7.5EV which according to the look up table in the Sekonic manual is 42 footcandles.

For those curious the softbox had a 500w tungsten bulb installed with two diffusion panels and was 134cm from front diffusion panel to grey card. By using knowing the footcandles and distance, I am able to come up with the “performance” of my own softbox and see how it fits in with lights by other manufacturers.

Here’s the quick reference I started putting together on different light “performance” numbers:

Diva Lite 400                  2808
Diva Lite 200                  1404
Arri 1k (med focus)        41,800
Arri 575w HMI (med)    50,000
Arri 200w HMI (med)   20,000
Lowel Rifa88 1000w      5,600
48″x32″ 500w softbox   850

Let me know if you think the test I conducted is accurate or not, I’d be interested to hear what people think.

Posted in Filmmaking | Tagged as: , , , | Leave a comment

Sony F3 time lapse sunset

Posted on October 14th, 2011

This is a short time lapse I did of a sunset over a lake using the interval function built into the Sony F3 video camera.

 

The time lapsed was about an hour with the F3 set to record 1 frame every 5 seconds. I started with the built in ND filter set to 1 (1/8th – which equals 3 stops of light) and f22 on a 35mm lens. As the histogram revealed the picture was getting darker, or less exposed, I started stopping down the lens ever so slightly throughout the time lapse. This did result in some exposure stepping which later in post I was able to smoothen out by overlapping those parts by about 4-6 frames. When I got very close to f1.4 on the lens I stopped recording, when up 3 stops to f4 and removed built in ND filter then started recording again.

Overall I feel it went well. The uncompressed full resolution version is absolutely amazing and what you’d expect from HD on the F3. In post I can blow up parts of the frame and the image still looks great. The footage seen here was treated with a curve and some post contrast to really bring out the colors.

Posted in Filmmaking | Tagged as: , , , , , | Leave a comment

Sony PMW-F3 PL mount removal

Posted on October 3rd, 2011

After having my own extreme difficulties removing the Sony PL Mount from the PMW-F3 (and not feeling good about having to tell people it was me doing it wrong) I thought an illustration showing exactly how to do it would be a good idea. It also would have been well appreciated in the user manual but is not located in it at all. 

Some caution on those “ears” located on the front of the Sony PL mount (three of them) and the ones on the rear ring (two of them) as they are made from plastic and can easily be broken off if you are thinking of tapping them or using excessive force.

I hope this helps many others out there who ran into difficulties like I did.

Posted in Filmmaking | Tagged as: , , , | Leave a comment

Sony F65 Sneak Peak at CSC

Posted on September 30th, 2011

It was a pleasure to attend the Canadian Society of Cinematographers meeting held at Videoscope in Toronto on September 27th, 2011 where I got to see the Sony F65 extremely up close and personal. (Yes I touched it!)

I had the opportunity to meet Ando (seen above) who is one of the Sony engineers flown in from Japan that worked on the development and design of the F65 along with several other 35mm digital cameras from the CineAlta line. The operational Sony F65 8k camera was equipped with a Sony SRW-R1 10-bit 4:4:4 (12 bit optional) recorder deck, and a tremendous Fujion 70-400 PL mount lens.

François Gauthier did a short presentation on the F65 camera and new 8k technology and explained how the new 8k sensor can create a better 4k image compared to conventional 4k CMOS sensors which use bayer-pattern sensors. While other manufacturers claim “4k” in fact the F65 accomplishes it by using dedicated photosites for each pixel instead of sharing some of them. 

Due to the high amount of data generated from the F65 (60 minutes of 4k @ 24fps = 1TB) Sony has developed a new open source codec which they call SRMASTER and there are already a wide variety of hardware devices that support it.

Just as important with any new codec, Sony has already started gaining support and commitment from a variety of vendors who will adapt the new codec into their software and hardware.

Some of the other impressive features with the F65 include a mechanical shutter to eliminate the “jello” effect with fast motion pans, wi-fi capabilities, and lastly as seen below, Ando demonstrates full control of the F65 with an iPad right down to adjusting the frame rate, shutter, exposure index, and start/stop recording!

Amazingly this camera is available in Canada for a pre-order price of only $85,000 if you order before December 23rd, 2011. To order, or get more information just email cinealta@sony.ca

The night ended with a variety of short clips shot on the F65 projected on a Sony 4k projector. The images were absolutely amazing, crisp, and free of any pixelation or anything you might look for to find a flaw in the image.

Posted in Filmmaking | Tagged as: , , , , | Leave a comment

Sony F3 SLOG Gemini 4:4:4 grading tests

Posted on September 22nd, 2011

Over on DVXuser.com Filippo Chiesa posted one of the World’s first official Convergent Design Gemini 4:4:4 recorder tests and was also kind enough to make available some of the images from the shoot in DPX format. The footage was shot on the Sony F3 with the S-LOG Gamma option and recorded via dual link to the Gemini recorder in 4:4:4. The actual video itself can be watched here: http://vimeo.com/29021459

S-LOG is not all that new and described in Sony’s 2009 white paper on S-LOG as a “digital negative” which allows maximum tonal reproduction of the camera’s imaging sensor. Since the F3 S-LOG gamma option allows you to enable 4:4:4 out via dual link it has raised a few questions on the true benefit of 4:4:4 over 4:2:2 color sampling. Some are calling 4:4:4 overkill and dismissing the Gemini recorder all together claiming that recording 10-bit 4:2:2 is “good enough”. Others however are certain the true benefits of 4:4:4 aren’t noticeable when simply viewing de-slogged footage, but instead during post production for grading, green screen, and visual F/X work.

One area I’m very interested in is to see how well S-LOG can hold up in its ability to recover detail in the highlights. I’ve been told that it’s WYSIWYG and that S-LOG only improves dynamic range and is not the same as a “RAW” image, but newer test results are showing us that there is in fact some possibility of recovering detail in highlights that may otherwise be lost if blown out. For me at least, this is one of the most powerful aspects of shooting S-LOG if it is the case because for decades when shooting video we’ve struggled to retain detail in highlights and finally now we may have it on a price friendly camera like the Sony F3.

Highlight detail recovery abilities aside, below are some sample grades I performed in post as a result of playing with Philip’s images. The first image shows the image in its flat S-LOG state and the rest show the progression of being de-S-Logged and then finally some subtle color correction and very minor grading. These images were downsized to 1/4 size and yet the detail still stands clear and beautiful.

S-LOG
Photobucket

DE-S-LOGGED
Photobucket

Grade #1
Photobucket

Grade #2
Photobucket

Posted in Filmmaking | Tagged as: , , , , , | Leave a comment

Sony PMW-F3 Super 35mm video camera

Posted on September 12th, 2011

In early 2008 the DSLR Revolution took us by storm, and in response largely to the DSLR revolution mainstream manufacturer giants such as Panasonic with the AF-100 and now Sony with the PWM-F3 are finally listening to the demands for larger sized sensor video cameras. (Or likely caught on after watching record breaking sales on Canon DSLRs.)

On November 8th 2010 Sony announced their new “affordable” Super 35mm digital camcorder to enter the line-up of existing high-end 35mm digital based cameras. Panasonic did introduce the AF-100 earlier in 2010, but it featured a micro 4/3rds sensor which only measures 18.9mm x 10.63mm in size while the F3 with its Super 35mm frame measures 24.89mm x 13.5mm.

Sony had yet another surprise when rumors surfaced that a prosumer version of the F3 was in development and would use the same Exmor CMOS sensor of the F3. The FS-100 was later announced as a prosumer option for recording Super 35mm digital images. One main difference is the F3 uses the XDCAM codec at 35mbit, while the FS-100 uses the AVCHD codec with a maximum bitrate of 28mbit.

Next to the RED line of cameras these cameras are part of the first wave of true Super 35mm video cameras to come close in competing with RED. Slowly we can expect Canon and JVC to follow suit which will be interesting to watch for over the next few months. Obviously RED has a larger frame of 4k and the ability to record using a RAW codec, however the workflow of XDCAM and 2k HD images has several post production advantages and can be really good for most applications. When you consider the option to work in 422 or 444 color space this is perfect for most situations.

One point of consideration is in order to fully utilize the 444 SLOG upgrade offered on the Sony F3, you’ll need an off-board recorder like the Gemeni 444 recorder which retails for $6000. Alternative options exist to record 422 to a variety of devices like the AJA Ki Pro mini, and newer devices coming out soon from Atomos and Sound Devices. To tally up the complete cost of the F3 to shoot SLOG excluding lenses and batteries:

Sony F3                               $14,000
444 SLOG Upgrade        $3,600
Gemini 444 recorder    $6,000

We’re looking at approximately $23,600 excluding any taxes and delivery. The Epic-S is likely just around the corner and Jim Jannard has already hinted at a couple surprises coming out before the end of this year. If the RED company holds true to what they have indicated in the past we could very well see a scaled down version of the Epic-X for a street price of $17k. If this ends up being the case then I see no reason why anyone would choose the Sony F3 over the RED, I certainly would not.

Posted in Filmmaking | Tagged as: , , , , , , | Leave a comment

Mixing 24p 25p 30p framerates

Posted on April 14th, 2011

Today you might easily find yourself in a situation where a client asks you to produce content in multiple formats such as PAL or NTSC. Or simply, the client has their own footage in one format but the camera you own is another. So the question really becomes how do we work with so many different formats, and what single format, if any format, is the best to use for distribution to all formats?

This article helps to answer some of that based on my personal experience as a digital cinema filmmaker since the early 90′s who prior to the advent of 24P capable cameras was getting motion-like film look using 50i long before it existed with the switch of a button on a camera. I will discuss some of the tools and methods I used which for a long time were my best kept secrets. (You read that correctly, 24p from 50i – yes!)

Let’s say your camera can shoot either format, PAL or NTSC . Which should you choose to shoot your content in? Traditionally people would tell you to “always chose the format of final distribution format”. Today, in 2011 that is becoming more and more meaningless. These days digital media delivery takes place in a variety of methods from computers, phones, media-set-top boxes, and a variety of other types of devices – all of which are format-proof since they can play practically any digital format and often the only limiting factor is the codec.

Prior to High Definition, Standard Definition PAL had a slightly higher resolution which offered some benefit if you were maybe looking at a potential film-out. In the digital world however since the aspect ratios between the two formats are different, adaptive pixel interpolation was needed to convert between the two which could result in some loss of image sharpness and could introduce ugly jaggies on edges. In the world of HD however, the resolution of PAL and NTSC are exactly the same making the only difference the frame rate.

How to interchange between PAL and NTSC? When you shoot 24p or 25p the speed is only 4% off from one other. Either one can ean be imported to the alternate timeline which effectively only alters the length of your project by 4%. Audio can then be squeezed or stretched by 4% easily without adjusting the pitch and you can use Adobe Audition, Pro Tools, or many audio editing programs will do this. For example if the video of your 5 minute production changes by 4%, the net effect is 12 seconds of time either way depending which way you convert. Most people will not notice any change in length of time since it would require them to know what the original time was to begin with. For example a band might notice on their music video, “Hey, what happened to 12 seconds of our song?”

What about 30p? Stay away from it unless you are absolutely sure you want it and you know why you want it. 30p is 20% off in speed from 24p or 25p and any attempt to interpolate footage to try and yield 24p or 25p from it is an absolute mess! If you think you want 30p for a less “juddery” filmlook but just can’t decide, shoot 60i since you can create 30p from it using field interpolation. You can even create good looking 24p from 60i using post production plugins. Should you decide you want 25p, shoot 60i anyway and use the 4% trick I mentioned to turn your 60i -> to 24p -> 25p. This still seems like a mess, but for the absolutely uncertain 60i is safe and also gives you a lot of options. (50i also can easily go to 25p and then 24p if needed.)

The bottom line would seem to indicate that the interlaced formats are the most flexible of all formats providing due to its ability of going to 30p, 24p and even 25p. (Remember that 50i can also be converted to 24p or 25p quite easily but will not go to 30p as well.) But if you’re going to shoot for a film motion-like look then stick with 24p or 25p and avoid 30p like the plague.

Posted in Filmmaking | Tagged as: , , , , | Leave a comment