Matching Vray light meter to real world lux meter?

[thebeals]

Hi everyone,

I am trying to replicate a simple scene that is illuminated by a single Ketra LED 40 degree spotlight. I used a light meter and found my lux value to be around 9,400 at the brightest part. I've photographed a few different objects including a macbeth chart and replicated in Maya to help ensure better accuracy.

I recreated the scene in Maya 2016. My scene is in millimeters. For the light, I utilized IES information for the same LED light. I am also using the vray physical camera matching my real camera's parameters. I scanned the scene with a geo scanner and matched it so I know my scene scale is accurate +- a centimeter. The entire scene is approximately 400MMX400MMX400MM.

When I get a render that is nearly identical to my linearlized photo of the scene (run through dcraw), my lux values (using Vray's light meter) are around 20,000 and not 9,400 like I was expecting. I am unable to match the lux values and have the render look like the photo. I am off by a factor of ~2.

I feel like I am missing something very obvious here. What am I doing wrong? Is there a scale factor I haven't considered?

Thanks for the help and clarification! I can provide a stripped down scene if needed.

[jasonhuang1115]

very interesting, thebeals. Can you post couple images to show the difference?

[vlado]

Can you post your scene? Or send it to vlado@chaosgroup.com so that I can check it out? There are quite a few numbers involved here so I can't say right away what may be wrong.

Best regards,
Vlado

[Rens]

This I suspect is a very tricky thing to get correct due to wavelength differences. LED lights have a very specific wavelength emission curve with strong peaks in certain areas. A light meter I assume measures all the light over the whole visible spectrum. As VRay adds R+G+B it might interpolate the rest different from what the LED actually emits? Just a guess.

[thebeals]

Vlado, I sent you an email with the scene. Thank you so much. You are the best!

I went back and measured the light again and found the brightest intensity is ~10,200 in real life and ~27,000 in Vray.

Here are some comparison images:


The first one is the photo, the second is my attempt to match it visually (~27,000 lux) and the third image is what it looks like when I try to match the lux values.

The photo and render match pretty closely. There are some subtle differences, but I think they are mostly due to lens differences such as vignetting and maybe due to some slight differences in light/camera placement.

@Rens, I was wondering the same thing. I was able to get a hold of an excel sheet of the spectral power distribution information a while ago for this specific light and it does have a roller coaster ride of peaks; although it is much better than most other LED's I have seen. Below is the graph if you are curious. The 6500k (what I have the light emulating) is pretty docile, so I'm not sure if it would still have a major effect or not. The 3500k value is crazy though; it would be interesting to turn the lamp to that value and see if it outputs any more energy to the light meter or not. If it did, then that could definitely be a major culprit!