Der Wannseepokal (PYC/VSaW) – DIE Frühjahrs-Jollenregatta für die FDs/Finn/420er und 29er – macht ja immer auf diesen Gewässern den Auftakt in die Berliner Segelsaison (obwohl die Optis eigentlich die Ersten sind :)). In dieser Jahreszeit kann es schonmal ordentlich blasen. Konnte man auch dieses Mal feststellen … mit Wind zwischen 4 und 7+ Bf in den April Graupel-Schauerzellen war auf jeden Fall viel Action drin und hier und da war das Material dem dann auch nicht mehr ganz gewachsen. So wurde es wiedermal ein starkes Event, mit vielen spektakulären Szenen die so manch einem nicht so schnell aus dem Kopf gehen werden ;). Sehr unterschiedliche sich sehr fix ändernde Licht- und Windverhältnisse – wenn nach den April-Schauern die Sonne herauskommt dann schneidend scharf und brilliant und klar. Schwer hier mit dem Weitwinkel etwas zu machen (in der Wettfahrt will man nicht so dicht an die Boote ran bei dem Speed), aber auch mit dem Tele wird es anstrengender, da auch auf einem Binnengewässer bei 7Bf etwas Boots-Bewegung ins Spiel kommt. So hats mir irgendwann den Lensmount-Lock an einer der D800 Bodies abgeschert – sowas ist ärgerlich, da man im Prinzip die Optik dann erstmal nicht mehr abnehmen kann unterwegs. Über den Wartungszugang, den die meisten Nikon Optiken haben, kann man allerdings mit einer spitzen Sonde später den abgescherten Riegel zurückschieben. Das Kameragehäuse ist allerdings nicht mehr zu gebrauchen und ein Fall für NPS, da der kaputte Riegel aus der Kamera fällt und sich dann keine Optik mehr verriegeln lässt. Das Ganze ist aber ein regelmäßiger Service-Fall bei Nikon – der kleine Stift ist eindeutig völlig underspecified für die großen Super-Teles und mit dem 400f2.8 passiert das oft.
There is a lot of confusion about the speed of MicroSD cards these days and various standards exist. Overall its easy to buy yourself into the wrong card type and for copter data acquisitions with the P4Pro the card speed is essential.
So here we go:
UHS-I and UHS-II (Ultra High Speed Classes):
UHS-II is the newest standard but is not supported by many yet. The Phantom series 3 and 4 all need UHS-I and definitely the fastest UHS-I cards. The UHS-I bus goes to 104MB/s whereas the UHS-II bus goes up to 300MB/s (theoretical limits – the cards will not perform at this speed).
UHS Speed classes are subdivided in to u1 and u3, while u3 performs minimum at 30MB/s write speed (needed for copter flights).
A new speed rating is called “video rating”. It scales from V6 to V90. The fastest cards are v60 right now (possible values are V6 V10 V30 V60 V90, but note that V60 and higher is usually UHS-II bus type and not supported by Phantom 3 and 4 series).
Speed Rating up to 10MB/s write – this is a slow class rating and c10 should be always possible for fast cards.
SDHC vers SDXC:
Up to 32GB capacity the cards have the label microSDHC whereas cards bigger (64-256GB) hold the label microSDXC.
For the Phantom3A (5MB/s max write speed) Phantom3Pro (60mb/s =7.5MB/s) and Phantom 4Pro (100mb/s=12.5 MB/s) series the fastest at writing to card seems to be right now the UHS-I SanDisk Extrem plus and Extreme pro cards (90MB/s sequential write). While this seems overkill the u3/V30 just certifies that you will never be below 30MB/s write speed under real world / all temperature conditions.
For some of the copter data processing folks Agisoft Photoscan turns out to be the most important tool/software to calculate point clouds, orthoimages and nadir data mosaics from copter photographs.
Problem: very long processing times with dense point cloud calculations with high or ultra-high settings (full resolution image matching with SfM (Structure from Motion) algorithms).
Some nice net finds show how multicore processing has its limits and why you should invest into GPU performance … and in high end 3D graphic cards.
Combining more than 20 CPU cores doesnt seem to speed up the process and combining more than 4 GPU systems also doesnt seem to help. There is only a minimal speed increase when you add more CPUs and or more GPUs when a 24 core system is already installed.
It boils down to a dedicated system with 2-4 Graphic cards with 3D acceleration (GTX-1080ti cards from Nvidea or if you can afford it a TESLA p100 based system), with approx 64-128 GB RAM and a dual i7 system setup.
Using DroneDeploy and the POI mode of DJI (this article was first published on jenacopterlabs.de).
Have been doing some leaf-off flights now in March on one of my favorite sites for complex tree crown point cloud mapping. To add to the leaf-on data from last autumn.
Fig.: Leaf-Off point cloud with combined height color coding and reflectance color coding.
This time I also checked the full automatic flight modi using the POI (Point Of Interest) mode from DJI and Drone Deploy with the P4Pro and the P3A.
Both work perfectly although I believe the POI mode can be dangerous when its done with low altitudes. You have to carefully check that the radius is free from obstacles when you define the center position, altitude and distance from center (radius) for the POI flight. When POI is started it begins with comparably low speeds. You can modify the speed setting when the POI mode started. Unfortunately you cannot modify this stetting before you start. Tuning the speed on an iPad or (even more difficult – on an iPhone) is kind of dangerous because you easily move the slide too much to higher speeds than wanted and the copter will immediately accelerate and start circling your POI object like crazy until you managed to move the slider to slower speeds. There is however always this stop button to kill the POI mode – this is handy and needed sometimes. When you press the stop button the copter stops the POI mode completely and waits for new commands.
The higher resolution version of the Schlachtensee Pano shot / panorama from early February. This was a magic week. Temperatures and light were just perfect and the ice was incredible untouched. This pano is from the P3A.
For those not familiar with the Berlin south west. The Schlachtensee region is known in Berlin as the area where the people who love dogs used to have some conflicts with those that go for a run and/or the other way around. I think the conflict has been cooling down now somewhat fortunately! In summer its great to have a swim here early in the morning. Skating in winter only works every 5-10years – so these days in February were kind of unique. (Pano from 4 different shots: resampled to 4300×1300).
The challenge clearly is A. to get the timing right (varies with the different swim styles/techniques), B. to be at the right position for the different styles, and C. get all the participants covered ;).
At the ISM C. is clearly impossible (there are just too many participants), timing works ok when you practice and/or when you can fire away with a speedy camera (but 9fps are not enough … ), B is difficult especially with the 50m and 100m competitions – there is basically not enough time with 50-100m runs. Freestyle seems to be the most challenging – its difficult to plan where the athletes take their breath – its a hit and miss – it feels as if you are always on the wrong side of the pool.
Some other issues come together here. The venue is very warm, and humidity is very high, its difficult to keep track where you are in the overall event program when you fully dive into shooting specific situations.
Some shots from different approaches that I tried:
The into the sun shots just capture a lot of flare and the sun star is very much Samyang 14mm like. This is an issue and it reduces the image acuity for 3D point cloud modeling for object measurements as well as for the landscape panorama approach.
The against the light shots look very much like the Samyang/Rokinon 14mm lens character. The sunstar has light rays that increase in size from the center and these rays cover the full frame of the 1inch sensor when the illumination comes from one of the outer edges. It clearly degrades the full image when the lens is stopped down to f.e. f8 but gets better when the lens is driven wide open. There seem to be some diffraction effects at work that are clearly not so great. You can drive the camera to generate sharp sun stars, but as usual here the characteristics of this star make a difference to some of us. The Samyang like star is not so well received. The linked flare issue however is much more a problem because it is hard to avoid when you have the sun in your frame and when the lens is stopped down to f8 or f11. For some scenarios a lens hood may be useful but its only functional for those shooting scenarios where the sun is not within your FoV.
To be honest I am bit puzzled … this is the first affordable 1-inch sensor platform with nice resolution and better DR, but – the flare issue is likely killing some of the potential ideas that you might develop with this machine. A pitty!
Comparison shot done with the Phantom 3A, clearly shows that the lens/sensor combi wont give the same amount of flare here (again different light level and different sun illumination angle):
Have done some tests with the new Phantom 4 Pro that I am using in 2017 for some research projects on photogrammetric point cloud mapping. This is a remarkable step forward in terms of image quality. The Phantom 4 Pro just creates much better image detail and acuity and resolution/sharpness is on a much higher level compared to the Phantom 3A. Its just giving back the DSLR like quality on tiny airborne systems. Also SNR (signal to noise) is much better and dynamic range ist enhanced. Only problem so far seems to be a slightly higher sensitivity to flare but I have to do some more analysis on this one. Its difficult to compare the flare issue because you cannot create exactly the same exposure / image concepts for different copters in the air. It will always be a wee bit different. So some ground based comparison needs to be done here.
This is a game changer for photographers that need an airborne system sometimes. As simple as that. It brings the image quality that we wanted back and you always can add HDR and Pano techniques to add from that. Great times to let your creativity fly!
Some shots from recent test flights with the P3A and the P4Pro: