First of all, probably you’d like to take a look on this unboxing video: https://youtu.be/KxscV0Fcu4A and, even better, on this other one that basically explain the same as I try to consider in this blog entry: https://youtu.be/fv3vt62VP7s
As you may know, a few minutes ago, new PXW-FS5 camcorder was announced, after one week of teasers with #NoLimits and #MissingPiece hashtags.
Despite it was just one week of rumours and shaking, and even more job done during last months (what a great team I belong to!), everything is finally disclosed now.
So, let’s talk about what was announced…
Many of you may have worked, or at least probably know the two models I’m going to especially refer to when talking about PXW-FS5: PXW-FS7 and NEX-FS700 (other models will show up eventually).
I use to comment that, from my point of view, those two camcorders belong to different generations: FS700 (like FS100) could be basically defined as “a box with a nice sensor”. Obviously, it’s much more than that, but related to ergonomics, that’s the first generation: priority was sensor quality, lens compatibility and camera features. But ergonomically, certain accessories were commonly needed in order to work in a traditional way, to hold the camera comfortably. And, as an example, LCD position was not totally proper for high takes (in some cases, users shot with the FS700 upside down, in order to monitor signal; image inversion was done in post).
However, evolution into bigger sensor cameras, with great features, but also with an optimized ergonomics kicked off with F-series (F5 and F55), and its modular body: they could be shoulder mounted, and any kind of accessory could be easily attached (follow focus, mattebox, steady…)
The one that probably meant a real revolution in this market, when referring to ergonomics, was PXW-FS7: it simply inherited the best from both worlds. On the one hand, it’s compact, agile (“run and gun”), with an easy operation both from “ascending” users from a FS700, like for “descending” people from an F55, and even for those ones that finally decided to replace their fancy EX3, EX1, 200… and on the other hand, it still keeps those stunning features we saw in F5, with a more “friendly” lens mount, with easier-to-find adaptors: the widely-known E-mount. Moreover; those lens that are typically used over E-mount (practically, ANY lens, since the flange back distance –sensor to mount end- is really short: just 18 mm), as they use to be light, which perfectly matches FS7.
No need to say FS7 was the desired gem for many users: flexible, powerful, carefully designed, robust… and even budget saving. But even at its –relatively- low price, some users prefer to keep working with, let’s say, “older” tools or workflows, either due to concept or even age, such as DSLRs, and they consider that working with a still camera that shoots video is “enough because it’s cheap”. They may be right, but we must also consider that camera body plus those accessories needed to make it a “fully working” video camera raises their budget even to the point to equal FS7’s price. So, that sentence saying “for the price of a FS7, two DSLRs can be bought” is not completely true.
So, ok, let’s assume that for some users, FS7’s price may still be a bit high. Or that a second camera could be a nice help for our production, but we cannot double budget. Or that there are certain situation in which FS7 may be “too much” (too bulky, too big…). For instance, an FS7 is obviously a multi-purpose camcorder, but when shooting a wedding, it’s worth having something even more agile. Or new producers for online media (YouTubers, Vimeo creators…): it’s not just a matter of placing a camera in front of them, but also to have a cinematographic look, with a shallow depth of field, yet not expensive. Or even a drone: putting an FS7 at risk is… too risky, and something lighter would be safer.
For all those situations in which an FS7 is not “reachable”, due to any reason, or in which something smaller would fit better, are the ones FS5 has been designed for.
Camera body, with no accessory attached, weighs only 0,8kg (core is magnesium built), and is totally modular. That’s why I was telling it wouldn’t be weird to se FS5 mounted in drones, small cranes, underwater housings… because size is also really compact, especially in comparison with the already not big FS7:
As in FS7, both screen and grip are detachable, so the camera can work without them fixed. This involves that in really tight environments (cars, lifts, closets, fridges…), we can shoot with the camera body and the controls (grip and LCD) only linked through a couple of cables, but not mechanically attached. So, camera can be lying in our legs or in some support while we operate from one side.
Moreover, even if FS7’s arm can be adapted (some spare parts are needed), grip is directly mounted over the camera body, and FS5’s center of gravity is in a very low position, slightly moved to the right, so it’s unlikely to “tilt” as if we had the arm more spaced from the body (it depends, logically, on the lens we use). Also, in high takes in which camera needs to be on an upper position (at very crowded events, like concerts, weddings, etc), our forearm will be less stressed than with classical handy camcorders.
No need to say that FS5’s grip allows menu navigation, and also blocking and unblocking its vertical angle, to set it in the correct position at every moment, so the camera can be fully operated just with the right hand, leaving your left hand to operate the lens (basically, focus, since if it’s a motorized lens, we can have zoom on the zoom lever, an iris/shutter on the grip dial).
SUPER35 MM SENSOR
What can be told about this sensor? To be honest, nothing that hasn’t be told before.
Maybe it’s worth reviewing a little of Sony’s recent history. You may remember one of the cameras that I personally was amazed about the sensor: PMW-F3. This imager was shared with NEX-FS100, and was incredibly sensitive. From my point of view, it was one of the milestones to jump into Super35 mm sensors in relatively “cheap” professional camcorders.
In fact, its behaviour was so, so good, that everything learned on it could be inherited in the well-desired 4K resolution sensors, keeping all the parameters, especially latitude/sensitivity. To illustrate it, FS5 works with about 14 stops. Simply amazing for an approx. 5k€ camcorder. And, yes, it’s the same sensor of that of a FS7 or F5 (be careful: I’m not talking about the processor – just the imager). Of course, FS5 allows to use S-Log2 and S-Log3 gamma curves; let me talk about that in the next section, apart from this one.
One of the most important features, asides from sensitivity or resolution, is the colour space that the imager is capable to capture. And when talking about FS5, it’s S-Gamut3 and S-Gamut3.cine.
And you may wonder why Super35 mm and not Full Frame size. Well, it’s a long-time discussion that might take several days, but let’s stick to: it’s the reference cinema sensor size. A Full Frame sensor is harder to “domesticate”; a much more careful (sometimes sick) job needs to be done over focus, and its aesthetics, yet “oniric”, may not seem totally natural. Sorry for being so subjective, but I wouldn’t like to offend anybody; I obviously don’t have the absolute truth on this.
Moreover, as I’ll tell when talking about the lens mount, we can “trim” sensor so that we can work with a size that is very similar to Super16 mm. Really “vintage”. J
S-LOG CURVES (just a technicalish approach)
I wouldn’t like to spend a lot of words on this (there are much more skilled people out there), but in our product sessions, we detect a lot of people that are really keen on S-Log gamma curves. And, of course, they are a great help, but we need to be careful with an aspect I’ll tell you later.
In a sensor, what is basically done is to convert light (input) into “output” (voltage, binary values…). Ideally, that transfer is linear: when 100% light that sensor is capable to acquire is reached, 100% of the output electronic value should be provided (those binary values, or voltage level).
However, nowadays we are working (as in FS5) with sensors with a huge latitude (let’s say dynamic range). So big that, even if sensor is not saturated yet, output is shown as overexposed, because it has reached its 100% maximum value (this is just a basic explanation; I wouldn’t like to jump into a conversation about if it should not be 100%).
If instead of applying a totally linear input/output transfer, we modify that curve (to logarithmic), so that highlights can be captured, and they are compressed, and later we apply an opposite curve (antilogarithmic/exponential) that, when multiplied by the other curve, results in a linear result, we could have got more dynamic range from the sensor (800%, 1300%, instead of the original 100%), without losing highlights, fitting all that range in that 100% output maximum level. Finally, as you can imagine, signal is linearized, with a much bigger range.
(Conceptually, it’s very similar to a RIAA preamplifier in audio: a turntable doesn’t provide a linear signal –“line”-, because it would need huge grooves to have all dynamic range recorded in its surface. A curve is applied that is corrected in the RIAA preamp or Phono preamp).
So, what is logarithmic gamma curve useful for? Most of us may have found situation in which a part of the image is overexposed, and we have simply accepted to lose that signal, or the low light signal, because we cannot host both levels in our camera: an indoor take with a window which light is passing through, clouds… and we simply kept that window overexposed, or clouds totally white. Now, we can recover all that info that was there, through the window on or all those grey levels in the cloud surfaces. In concept, it’s similar to that term that is becoming so fashionable in video world: HDR (High Dynamic Range).
Oh, and what is that warning I was talking about? When applying a curve (let’s say S-Log3), a minimum gain is applied (ISO3200 in FS5). This means that, in highlights, we’re going to be able to recover all that info, but in low lights, a default gain is used. If our sensor is not immune to noise enough, or it’s simply a noisy scene, when applying gain noise is also raised, and that’s why in low lights with S-Log sometimes some noisy areas appear. That’s why I recommend not using S-Log gamma curves for everything, but only if necessary; many people find it useful to use them always, as “info is there and I’ll recover later, no need to worry about overexposure”.
ELECTRONIC VARIABLE ND FILTER
Behind this long and strange name, a really useful brainchild is hidden.
If we look back, we can remember that FS100 didn’t have internal neutral density filters, since distance between sensor and lens mount made them really difficult –and expensive- to be implemented by those dates. In fact, that was one of the biggest improvements when FS700 was released: Sony could finally develop ND filters thin enough to fit in such a tight space.
In the late times, we could find a special type of filter on PXW-X180 and PXW-X160 camcorders, that apart from the “classical” step-by-step filter (1/4, 1/16…), had another kind that allowed light transmission to be modified in a much smoother way, with a more “analogic” behaviour, because up to 128 filter steps can be adjusted through a dial.
Well, now in FS5 we have, for the first time over our 4K Super35 mm sensor, that electronic variable ND filter.
Answer to “what is it useful for?”: Imagine (well, remember) that situation in which you were shooting indoors with your camera iris fully open, with a gorgeous DoF, and suddenly something changes in lighting (we may go outdoor, some window may be opened, we get into another room…); what’s the first reaction? We need to reduce light in our sensor, so we would probably close iris to prevent it from overexposure, but then our beloved shallow DoF becomes much larger. Then, let’s put some “sunglasses” on our camera –ND filters, actually. But we usually have only 3 steps, if they are discrete filters. Immediate effect #1: we will notice some sudden decrease of light in our footage. Effect #2: probably we won’t be blocking the exact quantity of light we really want to be blocked, so we may eventually need to modify any parameter like iris, or even adjust gain, or modify shutter. So, it’s clearly better to have ND filters with an “analogic” behaviour, with a smooth transition between steps. And, that way, we will reach a threshold in which light lying in the sensor will be approximately the same as we had indoor, preserving depth of field.
As written above, FS5’s mount is the already existing Sony “E mount”, that allows almost any lens to be installed, that has an internal interface to talk to the lens –for those that allow that- and that can host up to three kind of sensors: Full Frame (A7 and VG series), APS-C (NEX-EA50M) and Super35 mm (PXW-FS5, PXW-FS7, NEX-FS700, NEX-FS100). There is not only a huge range of E and A mount lenses (Sony LA-EA adaptor would be needed) available, but also an almost infinite combinations with third party lenses and adaptors.
So, if we are already EF lenses users, we would only need to bear in mind a crop factor that is close to 1.6 with Full Frame lenses (each lens will be more “tele”: an 80 mm will become a 128, so you need to be careful especially on the wide angle side), that we can even reduce to obtain 1-1.5 more stops through active adaptors (with a lens inside them).
In the opposite case, we could face that problem named “vignetting”: if we use very narrow lenses, we cannot cover the sensor totally. But in FS5’s case, from the release date there is a function to avoid that: “central scanning” in the sensor. Since it’s a Super35 4K-resolution sensor, we can just crop or use its central part, with 1/4 of the original surface (half height, half width) to have a sub-sensor equivalent to a Super16 mm, with all the consequences it involves (aesthetics, Super16 lenses, less rolling shutter derivative effects since there is less area and number of pixels to output…).
Personally, I think that E-mount is one of the best inventions that have been incorporated to the professional video market from the photo world in the last years.
SUPER SLOW MOTION (HIGH FRAMERATE)
One of the most loved FS700’s feature was its Super SloMo. Now, that feature has directly been integrated in FS5, with the very same high performance. It has one advantage and one disadvantage when compared to FS7.
The advantage is reaching up to 240 fps with a Full HD scan in the sensor (at higher framerates, less lines are scanned and signal is interpolated to “inflate” into 1080).
The disadvantage: as the camera works over a buffer, there is a maximum quantity of frames that can be stored. This means that, for instance, if we’re using a 24p codec and we use the 240 fps Super SloMo, we have just about 8 seconds real-time due to the buffer. As we are shooting at 10x faster speed, those 8 seconds would become a Super SloMo take that lasts around 80 seconds. So, for solving that issue with a time limitation in the buffer, we can not only shoot in a classical “Start-Stop” way (“Start Trigger”), but there are also two buffer-based modes: “End Trigger” and “Half Trigger”. If you wish to see how “End Trigger” mode works, you can take a look on this video, in which I demonstrate it with a FS700:
To sum up, that “End Trigger” mode allows us to… well… it may sound weird, but… we can “record the past”: as we are using a rolling buffer, there are at any time the last 8 seconds being stored on it. When we press the REC button, that buffer simply starts to download its recorded material since 8 seconds ago: this allows us to avoid being with our finger in the trigger waiting for something to happen (a football kick, a jump into the water…), but we simply wait for it to happen, and then we press the button. Simple and smart.
Oh, and let me drop you a little comment/advice: in the previous example (240 fps at 24p), we would be shooting at 10x, so we would be shuttering 10x faster, so sensor would collect just 1/10 original light. Thus, image is darker in SloMo, and if gain is applied (let’s say 9-10 dB to compensate, more or less), then noise is raised, and, logically, image can be more noisy than without Super SloMo. Please bear that in mind, because you may need a more careful illumination in your scene. 😉
WIRELESS AND WIRED CONNECTIONS
So far, it was not weird finding Sony cameras with NFC and WiFi: NFC (Near Field Connection/Communication) allows that, when a smartphone or tablet is brought near the camera, they “say hello” and a communication protocol starts, that will eventually be via WiFi. We could say that NFC would just setup a “handshake” between both devices, and they agree which “language” to use (WiFi).
This way, we can remotely operate our camcorder, and view (mainly just framing; obviously, some slight delay is experienced, and picture quality is just “enough”) from the appropriate application –Content Browser Mobile. As far as I remember, if it has not yet been installed in our devices, the process leads us to Google Play or Apple Store to download it without the need to search for it.
WiFi connection in FS5 also allows streaming directly from camera (no QoS), which is perfect for integrating such streaming inside a website or for sharing it within a LAN.
But everybody will have experience that situation in which tests have been done at home or at job, leaving everything configured and working fine… but when on field, there are hundreds of networks saturating the spectrum, or any frequency inhibitor, or some building wrought that seems to be a Faraday cage. Any inconvenience may be found, and that’s when we would rather have a wired connection. As you can see, FS5 allows that wired connection for streaming:
Since some models ago, Sony not only offers the classical “cold” shoes for accessories, but also an intelligent shoe, called “MI Shoe” (Multi-Interface Shoe) that allows communication between camera and connected compatible accessory to be established.
That way, we can connect an XLR box (XLR-K2M), a torch that can be commanded from FS5 (HVL-LBPC) or, which could be even more interesting, a wireless digital microphone receptor belonging to UWP-D series, that can insert audio inside the camera directly, with no external cable, via an accessory named SMAD-P3.
CLEAR IMAGE ZOOM
This technology can replace classical digital zoom (that is still also available) by a more evolved algorithm.
In a purely digital zoom, the way to interpolate is by an average calculation: if a value between a “1” and a “5” needs to be invented, let it be just “3”. Easy, but not optimal in many situations.
However, in Sony we realized that we have kinda “experience” in video world. This allowed us to use multiple analysis points due to the sensor resolution, and, according to them, find their parameters and compare to some classified image patters previously stored. For instance, when zooming in, most of the info is radially moving to the frame edges. Or, when panning, most of that info is moving horizontally. Or we can even classify some dark granulated areas surrounded by green as “a tree”. So, those behaviour patterns allow us to predict and interpolate much more accurately. And this algorithm in a branch, plus the classical digital zoom on the other path, provides a high accuracy and a way better picture quality when using zoom.
If interested, you can take a look on this video in which that technology is explained on a X70:
https://www.youtube.com/watch?v=fBJYWchFBtg (again, sorry for the video quality: it’s the best I could do at that moment L ).
This uses to be the most boring part to explain, because there is not too much to add to the info existing in the brochure or our website.
In Full HD resolution we have AVCHD codec up to 28 Mbps (1080/50p), like in FS700, and also XAVC Long GOP up to 50 Mbps (1080/50p, 4:2:2, 10-bit depth).
In 4K, it is still Long GOP, at a maximum speed of 100 Mbps (QFHD/25p, 30p, or the well-desired “cinematographic” 23.98p).
No need to say that, if you prefer to work with a different codec, you have HDMI and SDI outputs for your external recorder (very usual this days, especially if you have your workflow totally optimized, despite I personally recommend to test and taste XAVC codec if you still haven’t done 😉 ).
In XAVC Long GOP QFHD at 100 Mbps, a 64GB SD card can host up to 65 recording minutes. In Full HD resolution at 50 Mbps, it would be around 120 minutes in the same card. And in AVCHD at 28 Mbps, it would be 290 minutes.
And you may be wondering if there’s anything to be told about RAW. Yes, it will be capable to record RAW externally in the future, but we don’t know (and we couldn’t tell) when nor how as of today.
INDEPENDENT REC TRIGGER
It was still not mentioned, but we can record in SD or MS cards in PXW-FS5’s card “A”, while “B” card only allows SD cards to be used. So, as a difference to FS700, we have two card slots.
Also, we have three recording buttons: first one in the grip, second one in the handle, and third one in the camera body. So, you can imagine: we don’t only have the classical “Relay” mode (if A card capacity ends, recording continues automatically in card B without frame loss), and “Simultaneous” mode (same content in both cards; for instance, as a backup unit or to send one of them to editing suite and the other one for colour grading), but also a third mode, called “Independent”, in which we can independently command each card from a different REC button.
It is especially useful in events like live concerts, weddings, sports, conferences… in which it’s nice to have a whole footage in one card in just a few clips (in order to not lose any action, and for synchronizing a multi-camera production if any of the cameras has no timecode), and the other cards would be used for a regular clip-by-clip recording (each song, each stage, each speaker…).
Of course, no need to say that audio input are professional XLR connectors. But, as a difference to X70 or NX3, both are physically separated (we already saw that in FS700): one on the handle and another one in the rear part of the camera.
Audio recording is 2-channels; when recording in AVCHD codec, we can choose between Dolby Digital or –even better- linear PCM; in both cases, sampling frequency is 48 kHz and bit depth is 16-bit.
When using XAVC codec, audio is again recorded in LPCM at the same frequency, but in this case precision is stunning 24-bit.
Regular DC input (typical XDCAM series) can be used to feed the camera, but if you are a bit worried about FS5’s batteries, you’ll be able to keep your BP-U batteries, since it uses the same BP-U30, BP-U60, BP-U60T or BP-U90 that could be used on XDCAM handy or semi-shoulder models (EX1, EX1R, EX3, 200, X200, 150…).
That DC input I was commenting is also BC-U1 or BC-U2 battery charger’s output.
Probably I’m not the most adequate person to give a final sentence about FS5, not only because –obviously- I’m an interested party, but because I am not lucky enough to be a real user; I am just occasional, and less frequently than I would like.
However, there are some relevant aspects that are definitely clear for me:
- Ergonomics in FS5 are almost unrivaled by any existing camera nowadays; not even FS7
- Sensor is, simply, stunning. It allows us to use S-Log gamma curves, a crazy colour space, and we can reduce its size to Super16. For today, a sensor in a Sony camera is understood perfectly, and in cases like FS7/F5/FS5, totally demonstrated
- That chance to shoot in Super Slow Motion at up to 240 fps in Full HD is something that worked really well in FS700, and it has been fully inherited here
- Having the possibility to keep DoF (let’s be sincere: it’s one of the coolest features in today’s cameras) with its electronic variable ND filter is a very well designed and precise feature, to avoid overexposure in unexpected highlights
- Any lens can be used, without any renewal needed in our lenses collection; simply by using adaptors, we can assure that working with this camcorder is absolutely inexpensive (like that fact of using SD cards)
And, finally, little “details” like being capable to send streaming or files via wired connections, or prioritizing one detected face amongst others in order to follow it with automatic focus, or its focus magnifier, its MI shoe, its proxy recording, that independent recording mode over two different cards, the included codecs, the silent slow-spinning fan, that chance to buy just camera body or also with lens… are some features that make FS5 a very interesting camera, and I honestly think –and with- that it will be a real success. J
Kindest regards to all, and thank you very much indeed for reading this article. And sorry for its extension, but I hope it worth it.
Addition to the original article, on 12th September, in order to clarify:
SDI (just HD): 422 10 bit
HDMI (4K/HD): 422 8 bit
AVCHD (HD): 420 8bit, 28Mbps
XAVC-L (HD): 422 10bit, 50Mbps
XAVC-L (4K): 420 8bit, 100Mbps
ISO with SLog: 3200
ISO without SLog: 1000
Product Specialist / Content Creation
YouTube: Alvaro Ortiz