Hello,
I would definitely back up what everyone else has said, especially about starting with a "cheap" practice plane. Not only will that reduce startup costs and recurring costs/time that occur after a crash, but it is also far easier to transport and find a smaller space to fly in. Having to load up a big gas aircraft and its equipment just for a quick test of a new autopilot feature is a pain. I have built smaller and larger camera planes, and each have their advantages, but definitely nice to start smaller. The smaller electric aircraft also tend to have lower maintenance requirements. One thing to note, be wary of going with a very cheap plane. If you are pursuing a serious project, I think it is usually worth the extra money for a quality product that is more reliable and has better performance. Trying to save a bit of money in the short run often leads to more issues and frustration in the future, resulting in more fixing/debugging than flying.
I would also agree that purchasing will likely save money over designing/building your own (unless perhaps you are building in quantity and setting up a manufacturing process), especially for smaller ones. It also means you will likely be able to repeat the performance of an aircraft. One disadvantage is some may require extensive modification. Even small modifications can take some time.
We are currently using an electric Multiplex Mentor loaded a good 500-600+ grams of payload (not sure exactly at the moment). This is pushing the limits of the Mentor, and we don't have enough payload left for longer flight times. We typically only get 10-12mins on 3s/4400mAh to give you an idea. However, it is still small enough to hand launch and then recover in a very small clearing in remote forested areas. Landings are still very scary when trying to hit the spot given the heavy wing loading.
In the past I have used a Sig Mfg Rascal 110. This aircraft is a traditional balsa/plywood/film covering aircraft in a traditional high wing tractor configuration. We flew this aircraft up to a gross takeoff weight of around 12kg and 26cc gas engine. With careful vibration isolation on the engine and camera (nothing extreme) we were able to achieve very very good imagery and flight time of 1 hour on 700cc of fuel (the capacity can easily be increased, increasing fuel quantity is relatively easy because of the extremely high energy density of gasoline, adding a bit of weight really increases flight time, when compared to electric). We used a Nikon D3000 or D60 on this aircraft with a fast prime lens set at a fixed focus and a very very high shutter speed (this is how we beat the vibration issues...). There was also lots of other payload onboard including high powered wifi for 6km+ 18Mbit links to move the images down in near realtime, a gumstix computer to control the camera (with libgphoto2) and manage the datalink, and lots of vibration isolation/EMI/RF shielding and enclosures (one major disadvantage of gas, everything must be protected from vibration, even if you don't directly see the effects). However, this aircraft required several hundred feet of runway to takeoff and land, and landing occurred at around 70km/hr when fully loaded.
I have also seen the Senior Telemaster Plus used with good results as a larger workhorse. It can fly very slowly if lightly loaded - in a light wind you can almost land it vertically. It is also fairly inexpensive for a large aircraft. Be aware it is not the most sturdily built, to keep it light.
I haven't tried a glider type aircraft, but would really like to.
While a parachute seems like a good idea, one concern is weight. I have seen that typical parachute recovery systems may consume 15% - 20% of the total aircraft weight, which is very significant. Haven't tried any net recovery. Certainly agree that it adds complexity, and requires very careful design and test to implement properly. It will not be trivial.
One thing to consider is that if you end up with a decent smaller aircraft that can handle your payloads and handles your launch/recovery requirements, but simply doesn't have the endurance you are looking for, is you can consider simply accepting a lower endurance and planning your flight campaigns to get around this. For example, breaking a survey into several parts, and landing to swap out batteries in between. This is what we do with the Mentor. This only works if your range is small; having to fly a long ways away to reach the region of interest won't be helped by this approach.
Just my thoughts, perhaps they will be of use.
Thanks,
-Stephen Dwyer