If you had to shoot underexposed with this, how far under would you feel safe shooting at, knowing that you could bring those details back in post without minimal noise? One stop? One and a half?
Like Darr, I am also a commercial, production shooter. I simply wouldn't underexpose the shot - I'd light and expose it properly. No lights available? I'd get some then come back for the shot.
I don't spend time "learning my sensor" or "finding the limits" of the camera. I learn to trust my camera and it's meter, as well as learning the *meter* limitations. When shooting, I simply take a scene, meter for it properly, establish the shot, add/remove light as necessary, and produce the image. If the metered range is outside the spec, I'd add light/subtract light. I've used flashes, packs, flashlights, headlights, and even laptop screens to provide the light needed for getting the exposure right and in range for the camera as the scene requires. I've gelled whole windows with neutral density to reduce light as well as placed huge reflectors to add sunlight.
Fully manual, and what I was saying earlier was that if I had exposed the shot with a meter set to a time value of 1/60, then I presumed that the aperture reading it would've given me was around 1.4 (or a two stop difference from what I shot at). Saying this based on the fact that that is the how much I had to compensate in post to get it to the exposure level you see in both photos. Again, this was really just a casual test, hence no readings. When I shoot at home, I don't carry the light meter with me. Haha.
However, I am also a scientist, so I do my share of testing
when necessary to prepare for a shoot. I establish norms, perform experiments, and produce results so that I can draw proper conclusions. The biggest issue you have here in your "test" is that you don't have a proper metering to start with. The eye is easily fooled and a horrible light meter. I'd not call this a casual "test", but rather a casual "guess". You really need proper metering to establish your exposure. What your eye saw in the second shot as "probably close" is more likely many stops off due to that window.
I have both spot meters and color meters in addition to my camera meter. I trust my camera to produce results when it's within reason and expectation. When it's outside that range, time to get my spot out and investigate. Color accuracy is very important to my work, so I'll investigate lighting with the color meter until I'm satisfied that the light going into the scene is within spec.
When at home, I trust my camera. When on site, I trust my camera. When in studio...guess what, I trust my camera. If it produces weird results, I'll immediately work to establish whether it was the scene exceeding the camera's capabilities or an unexpected glitch. But, until I run into that, I don't worry about it because I trust my camera.
I typically do not trust my camera to expose a scene. 95% of the time I will meter manually (using a light meter).
You need to learn to trust your camera - that's how it becomes an extension of your arm/eye/brain and not a limitation. It's just a tool, and like any tool, you need to trust it. Look at any industry - construction, automotive, etc. and the people in it. If they can't trust a tool, they'll replace it with something that they can trust. Likewise, I'm a production shooter, so a camera I can't trust and understand is a hinderance and gets replaced. I'm not/never brand loyal - I could care less. I'm result loyal and as long as the tool is not in the way of producing those results, I'm all for it.
Also, once you learn to trust your camera, you'll immediately know when something is wrong or out of spec. Without that trust, you'll always be second guessing whether it's the camera or something else causing issues. You need that baseline and to get it, you practice (or test). If you prefer to test, you need hard evidence and solid technique with real measurements.
Also, this is interesting to hear about the Tiffen filters! I use this particular one for my video camera to take the 'edge' off and give it a more organic look, and wanted to see how it would play with this back. It's literally the lowest density you can get (1/8th), so it's no where near a stop of light, and I actually wanted it to shave off some contrast to see what the results would be. The white spot is a window, and there is a really thick curtain there that was serving as a diffuser (also cloudy outside).
How could one *not* know about the craptastic quality of Tiffen? It's been known, testing and retested, and passed around that Tiffen is one of the worst filters in the world for many, many years!
Yes, we know that white spot is a window - you're missing the point. Which is, that it's
very bright and 1) adding enough light that it's changing your perception of the correct exposure (since you didn't actually meter), 2) that it's
large and in-frame which is inducing flare, and 3) is changing the
contrast in the scene so that your shadows have even less detail which induces more noise.
There are instances where temperatures that are too cold also negatively impact the quality of an image, especially on CCDs. Have you experienced that on your singular, unnamed, back-up CCD before?
There are sooo many CCDs out there that are regularly chilled to the max. I'm talking liquid nitrogen or superconductive coolers that don't experience any of the artifacts you show. Like I've said, I've shot in sub-zero temperatures for hours with my CCD backs and never had a seam appear or color go patchy. Astro photographers chill their CCD cameras as well as many scientific (microscope) photographers. For astro work, the cooling reduces darkfield noise and prevents the CCD from heating when doing hours long exposures. No seams appear for them and color remains accurate - they are more critical about proper color imaging of planets than even studio photographers.
I've had my share of telescopes with CCD cameras that use Peltier chips to cool the sensors. My microscope cameras also have massive coolers built into them to keep the temperature of the chip chilled.
I've never seen temps too cold to affect CCDs. Ask NASA/JPL about how cold they run the CCD chips in space-borne instruments.