What Camera Should I Buy?
With my savings for new camera gear growing each month, it'll soon be time to decide exactly what to buy :)
I've been researching a number of camera and lens combos for a while and there are a few possibilities to choose from, each with some flaw and area of speciality.
Before I decide anything, I thought it would be nice to get the opinions of the people who have/will be funding the new gear :)
So my questions for you is, what do you think is most important?
1. Brute force resolution - e.g. 25k or even 32k HDRIs.
2. Low-light capabilities - cleaner night HDRIs with the stars; lower noise.
3. Shooting speed - Fewer artefacts from moving subjects, especially clouds and sunsets; more accurate and consistent lighting; shoot in places without attracting as much attention; less likely to miss opportunities.
Let me know using the form here: https://goo.gl/forms/bBHlfifrhsI3I8I12
Solving one of these issues is trivial, but solving two or even all of them at once is where things get complicated. So if I know what your priorities are, that'll help me decide what to go for.
For reference, this is my current gear:
Canon 600D (18 MP APS-C sensor), using Magic Lantern for exposure bracketing.
Canon 10-18mm f/4.5-5.6 lens (always shot at 10mm).
The limitations of my current gear are:
Maximum resolution - 16k (my current max) seems like a big number, but without dedicated backplates it's only high-res enough to stay sharp when using a 35mm camera in your 3d software rendering at 1080p. Zoom in any more and it becomes pixelated. I do however already shoot backplates fairly often when it's important, negating this issue.
Awful low-light quality - wonder why I don't shoot more nighttime HDRIs? This is why. It's a crop-sensor with a maximum aperture of f/4.5. Noise is everywhere.
Slow shooting speed - the 600D has a tiny buffer and slow processor, meaning sometimes ~5s delays between shots, giving time for the scene to change around me (e.g. clouds/people/light moving) even during consecutive exposure brackets. This can lead to artefacts, inconsistent lighting in the final image, and missed opportunities.
So clearly what we want is to solve some or all of those problems, without introducing any serious new ones.
Comments (12)
hi
Hi :)
Oooh waw, i didn't realise that, so it's only the resolution that really affects vram usage? In the live stitching video, if i remember correctly, you are not using EXR for the workflow? Is it now the case / a part of the ’new’ workflow you are developing? Did you found a trick to write metadata to EXRs files? What do you think of the hdri merging algorithm inside PtGui? Is it good enough to have accurate enough hdris? Is it ok to do all the : tiffs bracket merging + stitching + final export, inside of Ptgui? So what kind of fancy camera Philip has? Will the new workflow include the new equipment ?
Yep, resolution and bitdepth. I don't remember what I said in that video, but I don't need to write metadata to the EXRs (HDR can't store metadata either). I use RawTherapee to write the raw files to linear tiffs (which do have metadata), then merge those exposure brackets in blender to EXRs. I don't really trust PTGui's merge algorithm. I probably just don't know how to use it properly, but it gave me questionable results and makes the whole stitching process extremely slow since you're dealing with over 100 raw images instead of just a couple after merging. Philip has a Sony RX100 V, which can do 1/32000 and has a built-in ND filter.
Hi Greg, the new "Whipple Creek Gazebo" hdri is an 1046MB EXR, is it because you have created it with your "new" technique? It looks so good!
Not exactly - that is another one by <a href="https://hdrihaven.com/hdris/category/?a=Philip%20Modin" rel="nofollow noopener" target="_blank">Philip Modin</a>, who is using EXR. Personally I've decided to stick with HDR because EXR provides only negligible color differences in even the most extreme cases, while almost tripling file sizes. I am using EXR for most of my workflow though, and only converting to HDR right at the end. This means if I actually manage to find a real-world case where it makes a difference, I can just output EXR instead.
Personally, I am most interested (by a large margin) in low-light capability for more and better night HDRIs.
Whoops, skimmed too fast and didn't notice the form. Ha ha.
I'd suggest the Sony A7R III. It has great low light performance and it's Full Frame High Resolution Sensor means you can get high resolution HDRIs from much fewer total shots.
It's on my list of considerations, but from an HDRI perspective not much different from the a7R2 or even original a7R (which I already use for backplates). The main issue is how to shoot an arbitrary number of exposure brackets.
In my case, 16k is already too much: I'm using your pictures just to have a foreign background, most often than not with "Depth of field" on, two high-def characters and a 16k hdri map is the max that my 6gb video card can handle. Anyway: thank you so much for everything :-)
Good point, hardware capabilities is certainly a limit, but it's worth considering where the standard will be a few years from now.