Hi all.
Doing a harding test for the first time and getting fails, having hard time figuring out how to fix it. Getting Failed (luminance flash 1 incident / 11frames) and it tells me the frames it concerna and the luma issue on each fram but i cant find any apparent flashing? Its a camera move looping through a miniature world(real) with cg assets floating around, and the camera move is speeding upp. Any one thats has experience with hardintests and might whant to lend a guiding hand.
I’ve actually had to do alternate edits for UK due to “flashing.” If this is not an option, you might consider adding some motion blur across the frames that they call out.
Thanks. Tried applying a radial blur to simulate motionblur, to much thins happening in the image that normal motionblur didnt work. But still fail, now with even more frames wrong.
Try applying a colour correct and turn gain to 50%. Test the clip. It should pass, so have another go at 60%, rinse and repeat, raising the gain until it fails again. Don’t forget to include handles on the test clip, it could be the actual cut from one shot to another that’s triggering the alarm. At the end of the day, this is legal so you have to get it through, no matter how ugly the result… or have the client can arrange an alternate shot.
A useful website with advice for interpreting the results is… Cambridge Research Systems Ltd Whitedot Scientific Ltd - How To Interpret HardingFPA Results
How about adding a timewarp of 100 and with a mix value of 10-20? Maybe a bad idea for overall quality though.
It can be all sorts of things that set off the alarm.
I had a shots panning across products on a supermarket shelf that was triggering the test.
We had the machine so we could keep trying things and running the test.
Tried a lot of things and kept failing.
Fix was ugly. I had to reduce contrast and add directional blur.
try averaging the clips either side of the cuts, if there is no apparent flashing its probably the cuts doing it. Play with the colour, bring the gain down and highlights. Do what you have to, so it passes. Then work backwards to make it look the best it can. Not a nice place to be, especially if you need to deliver asap.
Thanks for all the great advice!
Really interesting this whole thing, After reading alot i still find myself clueless to why I’m faling. Done 6 attempts now following advice from above and still Fail and with 30£ a try i’m thinking i’ll just wait till monday and se what the support team on hardingtest site has to say.
On another note, as far as i can tell the video is tested using a physical “harding box” not software? Anyone know the cost of this magical machine? Assuming it’s super expensive.
Hi,
Yeah if you dont have access to it then it will get expensive and id imagine a pain to get it to pass as its a lot of trial and error. We are lucky to have it at work, sometimes i put out 10,20 clips to see how its failing. It used to be a piece of software that you bought a licence for to use. I believe its changed, so ill find out more tomorrow and let you know.
I found in the past i needed to locate the exact frame referenced in the fail report, then move a few frames either way whilst watching the scopes to see if anything moves or changes, or maybe even a tiny little luma spike. It might be a case of fix the frame rather than the shot.
How are you delivering the file? Is it getting made into a DPP?
Another thing we got to work once was using an inverted vignette.
The luminance flash 11 frames means that there’s high luminance for 11frames.
There’s something in the manual about how it calculates all of this. I can’t remember exactly.
However what you must do is lower the luminance level of that section. What I would suggest is using the colour correct, roll off the highlights on a gamma curve or something similar. Use the wavefor to guide you and make the Luma go under 80% for that section/shot. Or cover up the high luma area with something. Whatever looks least shit.
In general, it’s a pain in the hole.
Thanks all! Actually got it throught now 
Ended up stappeing frame by frame wacthing the scopes for the frames with highest change, found one, added a pretty harsh vignette on the frame with a small fade in/out and that seemed to solve it.
3 Likes
There are some cheaper tools now to do PSE testing. Editshare QScan is quite an affordable option if you need to do it regularly.
I have to do PSE fixes all the time and you get to know from experience the best way to fix it.
One trick I often use is to run the problem clip through an average then mix that back over the original very subtly. When it is a flashing sequence it often fixes it without being too visually obvious.
You can also drop the luminance or gamma of the highlights. If you can lessen/even out the changes between the frames then it is much less of an issue.