How to differentiate molten metal from molten salt when everything is glowing at 1050 C?

Hello, I am a molten salt researcher and my reactors run at ~1050 C. I can't look inside my reactor as it operates sealed, but I am attempting to differentiate metal droplets that may be floating on the surface from the molten salt using viewing ports, or briefly opening it up and using mirrors to observe from a distance.

The inside of my reactors glow orange at this temperature and as of now, even with them opened up, it is difficult to know if I am looking at anything interesting as they are bright. I hope to provide contrast and make it easier to see if anything is changing at the liquid surface.

Ideas I have:

1. Two opposing viewing ports, at an angle, cut into the sealed reactor so that I can observe the surface of the liquid. One port will have a light source shining into the reactor, and the other is a viewing port where I intend to filter out different colors of light using colored glass, and reduce the total amount of light coming through to enhance contrast. From what I understand, green light is most visible to the human eye and is far enough from the orange glow that I can safely try to filter all light that is not close to the green wavelengths. It seems reasonably easy to get a high power light source for reasonable prices (even just a laser that is incredibly out of focus).

How well this would work to create contrast, I am not sure, but I intend to take advantage of the metal's reflective nature. I have thought of blue light as well but notice that blue/black lights tend to have this fuzziness associated with it, likely making picking out details too hard for me.

1. I have a cheap digital camera and could attempt to remove the IR filter from the lens. Then filter out the visible light with multiple thin black trash bags which should allow a lot of IR through, but little visible light. Then I would add layers and layers of glass (should block IR) until I have achieved enough dimming to hopefully see some contrast between metal and salt. The idea being that they have different emissivity therefore radiating different amounts of IR. I am aware of filters but the budget is low so those may break the budget quickly if I'm not right about what wavelength light is best.

2. Welding goggles. While simple, I believe most of my light is closer to red-orange than blue-UV which I believe welding goggles tend to be more geared towards.

3. I can also attempt to briefly open the reactor mid-operation and use mirrors to observe it at a distance with whatever contrasting method that is suitable. I do believe this would be more difficult to see a clear image and also believe this may be an issue as white smoke can no longer be blown away as easily compared to when the reactor is sealed.

I am open to any other ideas as this is well outside of my breadth of expertise. The budget is fairly low (max 50-150 bucks for everything) as this is just an idea I want to try to make qualitative analysis easier and wouldn't necessarily get direct funding.