exposure issues during render

Apple Studio w/M1 Max
64gb memory
Resolve 19.1.4
Sequoia 15.4.1
Eizo CG2700S

I've had this issue occur sporadically over the course of time and it has taken place with various Resolve releases - There seems to be no pattern to the problem and it is very sporadic in nature. I shoot in Canon C-Log 3 pretty much exclusively.

The issue is this: sporadically the exposure changes within the entire clip during the render process - The last instance of this I fiddled with gamma and gain and ultimately adjusted the offset substantially - I assume you're not supposed to render already rendered files but this particular video took 4 or 5 attempts at re-renders before I could get a proper exposure - The exposure was being altered with each re-render so my adjustments were altering the exposure, it just took a few attempts before I was satisfied with the end result.

100% of my work is for internet display - I have used both gamma 2.4, and for awhile at this point I've been using gamma 2.2 which I've seen as the recommendation for internet display - To be clear these exposure changes are taking place prior to the file ever hitting the internet although I have experienced these exposure changes using either one of those gamma's.

The end result is the original edited file now looks too dark in terms of exposure when viewed on the monitor but the re-render makes it look good after exposure adjustments have been made thru hit and miss adjustments.

Maybe someone out there can help with this issue?

The only way I can imagine the exposure actually changing during the render is if you have an invisible (or hidden) dynamic level change hidden in the Timeline Node window, or if you have something similar as an Adjust Clip on the Edit page.

When in doubt, try a test with a temp timeline of just Color Bars. If it changes, you should be able to see it very quickly on scopes.

I have never seen this happen since post went digital in the 1990s, but were are some possible causes for level changes in the signal path in the analog days. Digital... it should be exactly one thing (like 1023 peak whites in 10-bit) and it shouldn't change even 1 unit the next day or a week later.