AVS Forum banner

621 - 640 of 833 Posts

·
Registered
Joined
·
860 Posts
got ya. in ur post it sounded like the offset variant was the reason, while the only reason the i1D3 would ever come closer to the K probes is ur custom settings.


so what are those settings ? can u share ? (no worries if not)
I did mention in the original post the i1d3 was using turbo settings, and in the post you quoted I said the settings were tweaked, but I can see how that could have been misinterpreted. To be honest, there isn't too much to share on this front as it would take experimentation per individual setup to find the optimal settings. It is just a case of adjusting the combination of Integration, Intelligent Integration and Extra delay until you get fast, repeatable measurements with no misreads. It can also differ depending on TPG too, and how that TPG is setup (direct connection vs network connection). I just set everything to minimum (for speed) and run a few small cube patch sequences, like a 5/6 point cube, check for mis reads and then raise whatever setting needs adjusting based on that.

do u happen to have stats that show the improved performance directly compared to a K probe ? (if u can share)
I have personally run the same tests multiple times to ensure it is 100% repeatable so as not to post misinformation. The results of the match compared over 1000 points to a Spectro did look "too good to be true" so I wanted to check it wasn't a fluke.

On average, the i1d3 takes 20 minutes to complete a 1000 point profile (Black patch Stabilisation set at 0.1)


With standard i1d3 settings, it takes 30+ minutes. The Klein can take between 15- 20 minutes depending on whether you have Avg Low Light disabled/enabled. My point was, the i1d3 can be significantly sped up to be comparable to Klein, and with the new Probe Match function, it is also significantly more accurate overall. The new probe match benefits i1d3 users more than Klein users because the Klein is already fast and is already very accurate when using FCMM.

This is Jeti vs i1d3 FCVM over the same 1000 point profile, but with patches below approximately 3 nits removed. It took the i1d3 less than 15 minutes to read 942 patches
 

·
Registered
Joined
·
2,194 Posts
When setting Extra Delay (which is critical for overall speed), a simple method is to use Stabilisation and set to a colour (Such as 100% Orange).
Then read a Grey Only profile and adjust the Extra Delay value until all grey measurements show no colour cast.

(We will be automating Extra Delay... but until then, this approach works well.)

Steve
 

·
Registered
Joined
·
1,081 Posts
To be honest, there isn't too much to share on this front as it would take experimentation per individual setup to find the optimal settings. It is just a case of adjusting the combination of Integration, Intelligent Integration and Extra delay until you get fast, repeatable measurements with no misreads.
thanks for sharing the pics. Tbh, CS users w/ i1D3 here would def benefit from ur exact settings, at least as a base as u say u've figured out to get the i1D3 to be more repeatable in low light yet not sacrificing too much speed.... that is the holy grail for the i1D3... :)

The new probe match benefits i1d3 users more than Klein users because the Klein is already fast and is already very accurate when using FCMM.
yeah, u do misunderstand this - hence the reason for my other post.

alternative offset methods benefit any probe, incl all K probes.

The probe's characteristics and the offset variant are two completely separate items as I said before. The Klein is not "already very accurate" w/ FCMM, b/c on WOLED this combination leads to less than ideal results.

Matter of fact the i1D3 - if configured to be very repeatable - w/ a better offset variant will outperform the K probes w/ FCMM on WOLEDs.

but naturally if both use the same offset variant the K probes will win... so the point is (in this example) that the better offset variant is the reason why profile data from any probe will ultimately provide more accurate data. The K probes need that too... FCMM on WOLED is a problem for all probes, not just the cheaper ones.

Getting the i1D3 to be more repeatable in low light (as in: enhancing the probe's characteristics) is def another big step, IF(!) that can actually be done to be close to the K probe performance... would love to see exact stats how u evaluated this...
 

·
Registered
Joined
·
860 Posts
yeah, u do misunderstand this - hence the reason for my other post.

alternative offset methods benefit any probe, incl all K probes.

The probe's characteristics and the offset variant are two completely separate items as I said before. The Klein is not "already very accurate" w/ FCMM, b/c on WOLED this combination leads to less than ideal results.
I'm sorry but no, I did not misunderstand anything. I never said it does not benefit the Klein, only the benefits weren't as substantial. And that is true. I saw the exact same thing in the recent past when comparing TMC (Bodnar) to FCMM in that FCMM worked very well with the Klein, but TMC improved things slightly. I posted my personal findings about this in the LightSpace thread some time ago

I am finding a very similar trend with the new FCVM method too


And when compared to FCMM


So yes, it will of course benefit all probes, but it benefits some more than others. I have never said anything other than that. The reason I showed the results for i1d3 probes because most users here will be using such a device.
 
  • Like
Reactions: mrtickleuk

·
Registered
Joined
·
1,081 Posts
So yes, it will of course benefit all probes, but it benefits some more than others. I have never said anything other than that. The reason I showed the results for i1d3 probes because most users here will be using such a device.
no, it cannot. it's an offset that is applied to a X|Y|Z value, none of this has anything to do w/ the probe itself.

u're mixing things up b/c u did two things, not just one, but u keep stating them as one

(1) u used a what u call a "new" probe offset method
(2) and u supposedly made the i1D3 provide more stable results in low light

re (1):
the offset method has the exact same effect for all probes - no extra benefit for any probe. absolutely impossible. by pure logic u must understand that. it only does a mathematical transformation on values.

re (2):
configuring the i1D3 to be more repeatable in low light (if u did that successfully and it can be tracked/proven), enhances the i1D3's performance, hence the probe, hence applying the offsets (no matter which variant, incl. FCMM) will lead to a more usable result. (yes: usable)

a heads up:
these offset variants aren't new - we've been using them forever (first discussed w/ Steve in 2012 but u know how that goes), and as I heard others have been using them at least since early 2019. They're now new to CS and they're definitely NOT unique to CS.

as u can see, the real deal here is - if that can be evaluated and proven - is the i1D3 settings that u say u found by stabilizing the probe at low light yet not sacrificing speed (which would be a problem on WOLED)...

like I said, would love to see direct stats against any K probe (in parallel evaluation) if u feel it now compares to them... I got no skin in the game re i1D3, but would love to see what u supposedly cooked up there... and like I said it would help a ton of CS i1D3 users...
 

·
Registered
Joined
·
2,194 Posts
Yes, FCVM will benefit some probes more than others, due mainly to the inclusion of a Y value as well as xy values for RGB.
(Note, FCMM probe matching is based on xy data for RGB, and Yxy for W only.)

So with probes that have filters that are not as 'accurate' as they could be, the use of FVCM will indeed offer a greater level of improvement vs. a probe that has more accurate filters, where FCMM will work better on that probe vs. the 'less accurate' filter probe.

Oh - and FCVM is totally new, and has NOT been available previously, on ANY calibration software.

Steve
 

·
Registered
Joined
·
860 Posts
no, it cannot. it's an offset that is applied to a X|Y|Z value, none of this has anything to do w/ the probe itself.

u're mixing things up b/c u did two things, not just one, but u keep stating them as one

(1) u used a what u call a "new" probe offset method
(2) and u supposedly made the i1D3 provide more stable results in low light

re (1):
the offset method has the exact same effect for all probes - no extra benefit for any probe. absolutely impossible. by pure logic u must understand that. it only does a mathematical transformation on values.

re (2):
configuring the i1D3 to be more repeatable in low light (if u did that successfully and it can be tracked/proven), enhances the i1D3's performance, hence the probe, hence applying the offsets (no matter which variant, incl. FCMM) will lead to a more usable result. (yes: usable)

a heads up:
these offset variants aren't new - we've been using them forever (first discussed w/ Steve in 2012 but u know how that goes), and as I heard others have been using them at least since early 2019. They're now new to CS and they're definitely NOT unique to CS.

as u can see, the real deal here is - if that can be evaluated and proven - is the i1D3 settings that u say u found by stabilizing the probe at low light yet not sacrificing speed (which would be a problem on WOLED)...

like I said, would love to see direct stats against any K probe (in parallel evaluation) if u feel it now compares to them... I got no skin in the game re i1D3, but would love to see what u supposedly cooked up there... and like I said it would help a ton of CS i1D3 users...
I guess we are talking at cross purposes here. My sole intention was to show that the new probe match method works, and works well. Nothing more, nothing less.

By "new probe match method" I mean it is new to me and any other general user of ColourSpace. Anything outside of that "fact" is nothing to do with me.

I can only go by, and show the results of my own measurements in relation to what I see with the Klein and the i1d3. That is what I have done and what I have commented on. Again, nothing more, nothing less.

i1d3 probe settings have been discussed many times in the past in the LightSpace thread. I haven't "found" anything. Just set everything at minimum (Int 0.25, Intelligent Int 1.0, Extra Delay 0). Run a small short profile and look for issues. Need to raise Extra Delay? Notch it up to 0.25 or to whatever value is required. Low light needs a bit more stability? Raise Intelligent Int to 1.5 or 2.0. When you have stable/repeatable reads, that will be your optimal settings. In my case, in my environment with my setup I can safely use Int 0.25, Intelligent Int 2.0, Extra Delay 0.25, and with those settings a 1000 patch sequence takes 20 minutes. With the Klein it takes 15-20 depending on whether Avg Low Light is Disabled/Enabled.

Of course, below a certain threshold the i1d3 will never be as "accurate" as the Klein. That is the limitation of the probe and no amount of setting tweaks will ever change that fact. But down to it's limits, if you tweak the settings based on your setup and environment then the i1d3 can read almost as fast as the Klein, at least when the Klein is used within ColourSpace as there are limited options to configure the Klein in ColourSpace. Extra Delay and Avg Low Light are the only two options available that directly affect the speed of the K10, whereas Int, Intelligent Int and Extra Delay can be configured for the i1d3.

As I said before, individuals will have to test for themselves. I have absolutely no idea why you think I have"cooked something up"?? It is nothing more than simple tweaking of settings as I have said multiple times.

It just feels like we are going around in circles as neither one of us is understanding what the other is trying to say. I am sorry if I am unable to express my meanings well enough, but I refuse to keep spinning on this merry-go-round repeating myself over and over. If you have doubts, you are of course free to test it all yourself.
 

·
Premium Member
Joined
·
9,823 Posts
I guess we are talking at cross purposes here. My sole intention was to show that the new probe match method works, and works well. Nothing more, nothing less.

By "new probe match method" I mean it is new to me and any other general user of ColourSpace. Anything outside of that "fact" is nothing to do with me.

I can only go by, and show the results of my own measurements in relation to what I see with the Klein and the i1d3. That is what I have done and what I have commented on. Again, nothing more, nothing less.

i1d3 probe settings have been discussed many times in the past in the LightSpace thread. I haven't "found" anything. Just set everything at minimum (Int 0.25, Intelligent Int 1.0, Extra Delay 0). Run a small short profile and look for issues. Need to raise Extra Delay? Notch it up to 0.25 or to whatever value is required. Low light needs a bit more stability? Raise Intelligent Int to 1.5 or 2.0. When you have stable/repeatable reads, that will be your optimal settings. In my case, in my environment with my setup I can safely use Int 0.25, Intelligent Int 2.0, Extra Delay 0.25, and with those settings a 1000 patch sequence takes 20 minutes. With the Klein it takes 15-20 depending on whether Avg Low Light is Disabled/Enabled.

Of course, below a certain threshold the i1d3 will never be as "accurate" as the Klein. That is the limitation of the probe and no amount of setting tweaks will ever change that fact. But down to it's limits, if you tweak the settings based on your setup and environment then the i1d3 can read almost as fast as the Klein, at least when the Klein is used within ColourSpace as there are limited options to configure the Klein in ColourSpace. Extra Delay and Avg Low Light are the only two options available that directly affect the speed of the K10, whereas Int, Intelligent Int and Extra Delay can be configured for the i1d3.

As I said before, individuals will have to test for themselves. I have absolutely no idea why you think I have"cooked something up"?? It is nothing more than simple tweaking of settings as I have said multiple times.

It just feels like we are going around in circles as neither one of us is understanding what the other is trying to say. I am sorry if I am unable to express my meanings well enough, but I refuse to keep spinning on this merry-go-round repeating myself over and over. If you have doubts, you are of course free to test it all yourself.
Thanks @liberator72, please could you share your 1,000pts sequence, or the settings you’ve used to generate it with Mike’s tools if not a standard one? I’d like to use the exact same patch sequence with the Discus to see if I could benefit from the i1d3 from a speed point of view when generating large LUTs, and keep the Discus for low light/contrast measurements. That would be a lot cheaper than getting a K10, assuming the the i1d3 is able to read the darkest patches of a 21pts LUTs without any issue (on LCOS, so with a black level as low as 0.0015cd/m2).
 

·
Registered
Joined
·
860 Posts
So with probes that have filters that are not as 'accurate' as they could be, the use of FVCM will indeed offer a greater level of improvement vs. a probe that has more accurate filters, when FCMM will work better on that prove vs. the 'less accurate' filter probe.

Steve
This is exactly what I was trying to say and show. It is a known fact that the i1d3 filters are not as accurate as more expensive probes.

A "rough" example of how FCVM benefits i1d3 users is this


So all I did here was record the four RGBW patches in ColourSpace and measured the patch sequence in ColourSpace which used FCVM. I then exported/imported the bpd's to LightSpace and read the exact same sequence with all the same settings except with LS using the traditional FCMM.

While it is not really a 100% direct comparison as it is effectively two completely separate sets of profile data, it is still using the exact same bpd data between the two, and is the exact same process I did for comparing FCVM and FCMM with the Klein shown a couple posts above, so should give a "rough" idea on the potential improvements.

Note: To make myself 100% clear, the image is NOT a direct comparison due to two seperate profiles being run, one in CS with FCVM and one in LS with FCMM. The profiles were run back to back with the same 10 minute Pre-Roll, same meter settings etc, so should be good enough to gauge the "approximate" amount of gain for an i1d3 when using FCVM over FCMM. I did this for my own benefit only, with no real intention of ever posting it due to the fact it is not identical profile data, but I kind of feel my hand has been forced into showing what I meant when I said it benefits the i1d3 more than it does the Klein.
 
  • Like
Reactions: mrtickleuk

·
Registered
Joined
·
860 Posts
Thanks @liberator72, please could you share your 1,000pts sequence, or the settings you’ve used to generate it with Mike’s tools if not a standard one? I’d like to use the exact same patch sequence with the Discus to see if I could benefit from the i1d3 from a speed point of view when generating large LUTs, and keep the Discus for low light/contrast measurements. That would be a lot cheaper than getting a K10, assuming the the i1d3 is able to read the darkest patches of a 21pts LUTs without any issue (on LCOS, so with a black level as low as 0.0015cd/m2).
The 1000 points used in the previous posts are just a generic Anisometric patch sequence used for measuring, comparing and evaluating the result of an applied probe match. The posts where you see 942 as "Total Points" is the same patch sequence with all patches below 3 nits manually removed (very quick and easy in Excel) to help with the Jeti reads when making the comparison. There is nothing special about them at all, they are just generic built in Anisometric patch sequences and are used only to verify and evaluate the accuracy of the probe match.

With FCVM, you just measure the exact same four RGBW patches (three times for averaging/stability) as before with FCMM. The only different thing I did when measuring those was to have Stabilisation activated (now part of probe matching) and set to 0.1 seconds. As always, meter setup/placement should be done correctly for best results.

I don't personally use his tools to generate patch sequences, I have a friend who has provided me with his custom sequences for actual profiling of displays. I cannot/will not share those as they are not my creation and I know he spends a LOT of time creating them.

Other sequences I sometimes use I generate using a script built by @bobof, but these are standard cube based sequences that can be optimised/sorted for drift etc on WOLED.
 

·
Premium Member
Joined
·
9,823 Posts
The 1000 points used in the previous posts are just a generic Anisometric patch sequence used for measuring, comparing and evaluating the result of an applied probe match. The posts where you see 942 as "Total Points" is the same patch sequence with all patches below 3 nits manually removed (very quick and easy in Excel) to help with the Jeti reads when making the comparison. There is nothing special about them at all, they are just generic built in Anisometric patch sequences and are used only to verify and evaluate the accuracy of the probe match.

With FCVM, you just measure the exact same four RGBW patches (three times for averaging/stability) as before with FCMM. The only different thing I did when measuring those was to have Stabilisation activated (now part of probe matching) and set to 0.1 seconds. As always, meter setup/placement should be done correctly for best results.

I don't personally use his tools to generate patch sequences, I have a friend who has provided me with his custom sequences for actual profiling of displays. I cannot/will not share those as they are not my creation and I know he spends a LOT of time creating them.

Other sequences I sometimes use I generate using a script built by @bobof, but these are standard cube based sequences that can be optimised/sorted for drift etc on WOLED.
Thanks, we all use custom patch sequences for calibration, that's not what I'm asking for. I'm asking how I can use exactly the same patches you did in your measurements. Standard anisometric doesn't mean anything. Depending on the patches, the speed will be faster or slower. So I can't compare the speed of my Discus to your optimised i1d3 speed if I don't use exactly the same patch sequence. I can use two different 1,000 patch sequences and get 30 minutes for one and 2 hours for the other, so saying "1,000 points" gives a rough idea, especially if you take out the darkest patches, but it still doesn't allow me to use the same sequence to compare speed between my Discus and your optimized i1d3.

It's fine if you don't want to share your patch sequence or the settings you use, or don't want to use a standard patch sequence and provide timings for that, but for comparative purposes the same patch needs to be used to compare apples to apples. If you can report times using a standard cube (at least 10x10x10), that would be fine too.

Thanks again for sharing your results, it's very useful to other i1d3 users.

EDIT: I fully understand how FCVM works and the difference with FCMM, I am purely interested in your speed optimisation to see if the i1d3 could complement my Discus or not.
 

·
Registered
Joined
·
2,194 Posts
Standard anisometric doesn't mean anything.
Yes, it means the default 10^3 patch set generated with ColourSpace, with the 'Anisometric' option set (which is the default).

Steve
 

·
Registered
Joined
·
1,081 Posts
I have absolutely no idea why you think I have"cooked something up"??
:):):)

because u keep saying repeatedly that w/ ur i1D3 settings u now match K probe performance... here was me hoping.... :rolleyes:

but after clearing this up, I doubt this comes close to K probes or CR100/200, but I'm sure it improves the i1D3 and that's obviously good...

btw, it's not just the total time period it takes to finish the profile, testing for "improved repeatability" (due to revised probe handling) takes more than reading and adjusting until u get one somewhat stable result...
 

·
Registered
Joined
·
860 Posts
Thanks, we all use custom patch sequences for calibration, that's not what I'm asking for. I'm asking how I can use exactly the same patches you did in your measurements. Standard anisometric doesn't mean anything. Depending on the patches, the speed will be faster or slower. So I can't compare the speed of my Discus to your optimised i1d3 speed if I don't use exactly the same patch sequence. I can use two different 1,000 patch sequences and get 30 minutes for one and 2 hours for the other, so saying "1,000 points" gives a rough idea, especially if you take out the darkest patches, but it still doesn't allow me to use the same sequence to compare speed between my Discus and your optimized i1d3.

It's fine if you don't want to share your patch sequence or the settings you use, or don't want to use a standard patch sequence and provide timings for that, but for comparative purposes the same patch needs to be used to compare apples to apples.

Thanks again for sharing your results, it's very useful to other i1d3 users.
As stated, it is a standard Anisometric patch sequence. By that, I mean just set 10 point cube within ColourSpace, and select Anisometric. The patch sequence used for the comparison is built in, not custom (with the obvious exception of removing low light). Sorry, I assumed that was a given.

But yeah, it is just a standard sequence you can select with the sliders in software, but I've attached it anyway, along with the sequence with below 3 nits edited out.
 

Attachments

·
Premium Member
Joined
·
9,823 Posts
As stated, it is a standard Anisometric patch sequence. By that, I mean just set 10 point cube within ColourSpace, and select Anisometric. The patch sequence used for the comparison is built in, not custom (with the obvious exception of removing low light). Sorry, I assumed that was a given.

But yeah, it is just a standard sequence you can select with the sliders in software, but I've attached it anyway, along with the sequence with below 3 nits edited out.
Nothing is a given :)

But thanks, that's the confirmation I needed to have, that you were using the standard internal 10x10x10 cube with the anisometric option enabled. [EDIT: apologies, re-reading your initial post, you had provided this information already].

Thanks for the below 3nits removed set. As my Discus reads much lower than the i1d3 it will be fine reading these, but that will help for the i1pro to compare accuracy (like it did with your Jeti), so I'll use that with the spectro.
 

·
Registered
Joined
·
1,081 Posts
Oh - and FCVM is totally new, and has NOT been available previously, on ANY calibration software.
the problem w/ this statement is obviously...

(1) u only know what u have access to, there's a billion other solutions u don't have access to that have features that LS/CS do NOT have
(2) 99% of the reason why some new stuff ends up in LS/CS is b/c users tell u repeatedly (an estimated ten thousand times) to implement it - u fighting/arguing like a madman yet then implementing it and presenting it as ur own idea and as a genuine CS feature :D:D:D

therefore LS/CS is often the last place where stuff finally arrives, and then it takes a while until it works... if it ever works properly :rolleyes:

so everything is "unique" and "totally new" to CS same as LS updates were lifelong.... :D

So with probes that have filters that are not as 'accurate' as they could be, the use of FVCM will indeed offer a greater level of improvement vs. a probe that has more accurate filters, where FCMM will work better on that probe vs. the 'less accurate' filter probe.

while this is obviously true on paper, none of this matters much if the probe is not stable - hence back to square one, as I was saying before. this is the reason why I was asking why he kept stating it comes close to the K probes in low light...
 

·
Registered
Joined
·
860 Posts
:):):)

because u keep saying repeatedly that w/ ur i1D3 settings u now match K probe performance... here was me hoping.... :rolleyes:

but after clearing this up, I doubt this comes close to K probes or CR100/200, but I'm sure it improves the i1D3 and that's obviously good...

btw, it's not just the total time period it takes to finish the profile, testing for "improved repeatability" (due to revised probe handling) takes more than reading and adjusting until u get one somewhat stable result...
Again, we are obviously either talking at cross purposes or completely misunderstanding each other. I have only said that it makes the i1d3 almost match the Klein for speed (which it does with a 250 ms read time for i1d3 above 2 nits) and that FCVM makes the i1d3 far more accurate than FCMM when compared to the Spectro (which it does as shown).

And what makes you think I haven't tested the stability and repeatability of the i1d3 with those settings properly? I have, of course! The quick fire method I mentioned before is just an idea for others to try, but I have repeatedly said it is up to individuals to experiment and test for themselves in their own setup and environment to get the best results.

It's Sunday, I don't intend to spend all day here debating things that aren't debateable. Of course the Klein and CR probes will ALWAYS outperform a £180 consumer probe. The fact the cheap probe can be customised to perform admirably against a £6000+ reference probe should be welcomed, not cast off as poppycock :rolleyes:

Anyway, I'm out. It works for me, feel free to test it for yourself to see if it works or you too.


while this is obviously true on paper, none of this matters much if the probe is not stable - hence back to square one, as I was saying before. this is the reason why I was asking why he kept stating it comes close to the K probes in low light...
I see your edit after the fact..I have NEVER said the i1d3 matches the Klein probe in low light. I said explicitly that the i1d3 almost matched the Klein for speed over the course of the profile (which it absolutely did!) and was improved in accuracy due to the FCVM probe match. I later clarified why the i1d3 was faster than usual, by adjusting Integration to 0.25. And that Intelligent Int was 2 nits. How are you not understanding this???

It just seems you are trying to pick an argument with me for the sake of it, and I have better things to do with my time. Have a nice day
 

·
Registered
Joined
·
2,194 Posts
Just to reiterate, FCVM is totally new, and has NEVER been available in previous releases of any calibration system.
It has not been discussed previously, and only recently did we mention what we were working on with with the Alpha Test team.

Steve
 

·
aka jfinnie
Joined
·
4,219 Posts
If I recall correct LS / CS standard patch sequences have a 1.1 power curve (so a slight bias to darker values) and are rounded down to the nearest 8 bit integer (if you were into trying to make your own sets for some purpose).

The patch ordering stuff @liberator72 mentions is only useful to OLED owners really, and attempts to order the patch sets as best as possible so that the WRGB pixel utilisation and total power are as close to constant over a rolling period as possible. The built in ansiometric in LS / CS is pretty good in this regard (was originally not very...). The sequences I generated seem to perform a little better if you're watching the panel drift results, though I'm not sure there is any conclusive evidence that they result in benefit in the measured profile (though folk seem to like them). I don't actually use them myself as they don't serve any purpose on JVC PJ.

FWIW the tool for generating those sequences is freely downloadable. I did upload it previously to a different thread. I've added the link to my sig as I'm tired of trying to find it myself! :)
 
621 - 640 of 833 Posts
Top