I have seen a lot of talk around what to buy with regards to the Inspire 1 Pro, A Phantom 4 Pro or an Inspire 2 with the X4S or even the X5S, it’s a hard decision especially if you already have an Inspire 1 so I thought I would go into some detail around the differences with these cameras and the reasons to go with what model.
Let’s start with the Inspire 1 Pro – X5
Released in September 2015 this was the first of the home grown big cameras for the DJI ready to fly platforms, fitted with the same Micro Four Thirds sensor as the Panasonic GH4 this was a large step from where we were with the X3 and Phantom 3 Pro, with a wide range of interchangeable lens options and large 16MP what was there not to like but once people started to get their hands on it some of the limitation of the SOC and codec became apparent especially in 4K, most of this is because at this point DJI were still limited to using the same Ambarella A9 SOC that they had used on the X3 and Phantom 3 so even though there was this massive new sensor they were still having to compress all that extra data into the same amount of space and codec as before, the same limited 60mbps data rate and H.264 codec with the GOP structure, this resulted in some people seeing flickering, moir and just general compression issues with the X5 footage.
Looking at the X5 as a still camera only it performs extremely well but improvements and benefits over the X3 were mostly because of the extra DR and lens options, it was held back compared to the GH4 that it shared its sensor with because of the SOC sadly.
The X5R having the same sensor as the X5 but DJI got around the SOC limitations by going completely around it altogether and going to full DNG RAW, they moved all the image processing out of the camera into a new upper section that also accepted the propriety SSD drive to store the massive DNG footage on, this was a moon size leap forward and the X5R is an outstanding camera and performs extremely well in all aspects, while it still used the A9 SOC for the SD card footage the SSD RAW DNG is far beyond anything that had come before it, the only real complaints with the X5R is its large size and weight pulling flight times to around 9 minutes in the Inspire 1, also the time it takes to extract the RAW footage from the SSD to your computer via the HUB can be painful as well.
The Phantom 4 Pro and Inspire 2 with the X4S
Both the Phantom 4 Pro and the X4S camera for the Inspire 2 feature the same 1″ 20MP Sony IMX183 Exmor R sensor coupled to a low distortion auto focus lens offering a 24mm equivalent FOC, it has a maximum ISO of 12,800 and dynamic range is increased by 1 stop over the X3.
Further to this new sensor DJI is the first to use the new Ambarella H1 SOC and it brings some massive benefits.
The H1 offers some big improvements over the A9 and the most notable is it supports recording at 100Mbps in all codecs in 4K at up to 60fps, this is over a 60% increase in bitrate compared to the X5 and X3 and is a hugh amount of extra available space to compress the sensor data into.
This is a massive improvement over anything we have had before on a RTF other than the X5R and brings the camera inline with things like the RX100 and GH4.
Next there is a new codec – HVEC
The next change that came with the H1 is H.265 support, the problem is H.264 was originally developed in a 1080P world and was never designed to compress 4K video, many of the complaints with all the A9 cameras so Gopro, the X3 and X5 is the highly compressed output.
H.265 was specifically developed with high resolutions in mind, as part of the compassion algorithm both H.264 and H.265 both uses the same three frame types, I-, B- and P-frames within a group of pictures, this is known as the GOP structure and looking more closely there are a number of specific elements that have been improved with the H.265
The first is Macroblocks, this is the method the codec uses of splitting the image up into sections to be compressed, the codecs job is the take the raw sensor data and try and compress that into as small a space as possible to fit on to a sd card, to do this it it basically cheats a little, It takes one full new frame with all the data in then for the next frame it examines it and basically looks at parts that have no or little change compared to the first one, then instead of storing all the same data every time a new frame is created it looks at what has not changed since the first one and basically creates a link to that section rather than store the actual sensor data, when that frame is recreated in playback it takes part of the data from that first frame in the sequence that its been linked to and only puts in the bits thats changed since that was taken, this process is called the GOP structure and usually in H.264 its every 8 or so frames it will actually reset and place in a whole new first frame in then start the process again, basically this whole process means the file size is much smaller while trying to retain as much detail as possible.
While this is a very basic explanation of what happens it gives you an idea of the process, here is where the changes are between the two codecs, H.264 uses macroblocks up to 16×16 to spit the image up for encoding, H.265 on the other hand divides pictures into what’s known as “coding tree blocks” (CTBs), this is an area size of 64×64 then depending on the image and settings the size of the CTB can be 32×32, 16×16, 8×8 and 4×4, this means in the areas of little change it can use larger 4×4 blocks but in high detail sections it can go up to 32×32 to preserve as much of the information as possible, this means in real life use that the codec needs less space to store the information in areas of little change leave more for areas of high detail, overall it can retain detail better at higher resolutions like 4K.
H.264 Macroblocks vs H.265 CBT’d
Further to this H.265 has a large number or further enhancements including better variable-block-size segmentation, improved deblocking and motion compensation filters, sample adaptive offset filtering, and better motion vector prediction and precision.
But what does all this mean for you ?
Well basically the data from the sensors is being compressed in more efficient and improved way and more of the actual sensor data that matters will be retained in the footage, you should see more dynamic range, less moir and an overall improvement in output compared to H.264 at the same data rate.
So should i use H.265 over H.264 all the time ?
Well that depends on a few things, it’s not nearly as compatible with existing playback devices as H.264 is, many devices have dedicated hardware for decoding H.264 while equivalent hardware for decoding H.265 is significantly less common, there is limited hardware decoding support in computers as well with only the 7th generation “Skylake” Intel processors fully supporting hardware decoding, overall working with H.265 right now is more time consuming and harder so while it has its benefits you may not want to use it all the time, but what it does do if give you more options.
Either way even in H.264 these cameras with 100Mbps and 1” sensors will offer some big improvements over both the X3/P3 series and the X5 in video recording.
The X5S is the next step in DJI’s larger camera line, fully compatible with the same lenses as the X5 and X5R its whole new camera with some major changes.
With both the X4S and X5S in the Inspire 2 the image processing has been moved into the craft body, DJI took its first steps with this with the X5R moving it into a larger upper section of the camera with the SSD sticking out the side, the problem with this is it added more weighting meant the camera was very expensive and any future upgrades would need all new processing again, With the Inspire 2 DJI went towards what they did with the phantom series and moved the processing into the craft body, DJI call this the Cine-Core 2.0, this unit contains all of the image processing, the Lightbridge radio and the main flight controller and IMU.
At this moment in time there is no information on what processing is in the core but you can assume it follows similar setup to the X5R with the SD card footage being handled by its own SOC likely to be the H1 and the SSD footage using dedicated image processing.
The upside to this is a far cleaner craft with all the processing onboard but this does mean that the cameras of the Inspire 1 and Inspire 2 are not interchangeable.
On the X5S sensor is self very little is known but If rumours are to be believed its could be using the same M43 20MP sensor thats in the new Panasonic GH5, its using a a fairly unusual resolution (5280 x 3956) and is capable of recording in 5.2K, neither of these are common some believe it could imply DJI may be using a multi aspect sensor.
Just like the X5R the X5S footage is recorded to an SSD drive with proxy going to the SD card, the SSD is now located in the rear of the Inspire 2, with the new Cine-Core 2.0 DJI have added both ProRes and Cinema DNG but unlike the X5R licences are need to use these, the X5R no licence was required but the cost of the X5S camera is substantially less to reflect both the processing being onboard the craft and the necessary licences.
So what camera do you choose especially if you already have an Inspire 1 ?
It comes down to a few simple questions regardless of what craft you currently have.
Do you need interchanges lenses ?
Do you need duel ops ?
If its yes to any of these then its got to be the Inspire 2, other wise go for the Phantom 4 Pro, even if you have the Inspire 1 looking at this from purely video point of view the X5 sensor is substantially larger than the 20MP 1″ sensor of the Phantom 4 Pro and X4S the problem is the X5 has no where to stick all that information for video, the X5 is just not worth a look unless your specifically doing still only, the X5R is an option but is still very expensive and is actually the same price as the Inspire 2 craft only.