Cards Mold/"Tina" Alibaba Ceramics Master Consolidated Info Thread (4 Viewers)

[SeanGecko]

Not sure if done before, but trying to predict Tina printer error for chips. Took the images of the digital color wheels in the threads as well as the actual printed versions.. put side by side.. top is digital, bottom is output.. from past pages in thread.

View attachment 1337001

Then reduced size (get average color vs grain) and recorded CMYK values for digital (input) vs actual (output) - Link to file below for reference:
https://docs.google.com/spreadsheet...ouid=112612925856116927892&rtpof=true&sd=true

View attachment 1337004

Then did a regression analysis against all to get a predictor equation for each CMYK value (C, M, Y, K tabs - did not go through p-value exclusion loop, e.g. omit >.15 values and rerun):
View attachment 1337005

That resulted in an overall color prediction:
View attachment 1337007

Bottom line... on average predictor equation from regression would forecast the true CMYK output 10 total points (C+Y+M+K) closer than the input CMYK value... but feel like there is opportunity to do a polynomial/differential look to go further because some of the correlations don't appear linear... e.g. magenta:

Magenta digital vs magenta actual, across just pink/purple chips (highest magenta):

View attachment 1337011
Edit: I like this example because it also clearly shows magenta over-indexes in actual printing.. have seen a few times in chips before.

Any thoughts? I'm trying to think through a way to run a better statistical analysis (non-linear, multivariable) to get a predictor equation without doing too much manual creation of color plots/fitting.

If we can well correlate... my thoughts are we should be able to get good faux inlay results... assuming Tina's "error" for each color is somewhat consistent, but may need a similar analysis also done on the inlay printing as well (assuming separate printer for stickers/inlays vs chips).

That's a lot of words

 

[Colquhoun]

Not sure if done before, but trying to predict Tina printer error for chips. Took the images of the digital color wheels in the threads as well as the actual printed versions.. put side by side.. top is digital, bottom is output.. from past pages in thread.

View attachment 1337001

Then reduced size (get average color vs grain) and recorded CMYK values for digital (input) vs actual (output) - Link to file below for reference:
https://docs.google.com/spreadsheet...ouid=112612925856116927892&rtpof=true&sd=true

View attachment 1337004

Then did a regression analysis against all to get a predictor equation for each CMYK value (C, M, Y, K tabs - did not go through p-value exclusion loop, e.g. omit >.15 values and rerun):
View attachment 1337005

That resulted in an overall color prediction:
View attachment 1337007

Bottom line... on average predictor equation from regression would forecast the true CMYK output 10 total points (C+Y+M+K) closer than the input CMYK value... but feel like there is opportunity to do a polynomial/differential look to go further because some of the correlations don't appear linear... e.g. magenta:

Magenta digital vs magenta actual, across just pink/purple chips (highest magenta):

View attachment 1337011
Edit: I like this example because it also clearly shows magenta over-indexes in actual printing.. have seen a few times in chips before.

Any thoughts? I'm trying to think through a way to run a better statistical analysis (non-linear, multivariable) to get a predictor equation without doing too much manual creation of color plots/fitting.

If we can well correlate... my thoughts are we should be able to get good faux inlay results... assuming Tina's "error" for each color is somewhat consistent, but may need a similar analysis also done on the inlay printing as well (assuming separate printer for stickers/inlays vs chips).

Keep in mind that while we set up our files in CMYK as requested, it’s unlikely that Tina is using CMYK printing. It is more likely 6-color printing, which would explain her ability to get very bright colors, especially in the blue range… and would also explain why certain colors print more as expected (red and greens) and some are a crapshoot (blues and purples).
Also, I would be careful when comparing the art to other members’ photos. The phone tends to brighten and saturate colors and they are often not accurate to the true chip color. I’ve had some Tina samples that looked very different when holding them, but look almost identical when photographed with my phone.

 

[chamber2skipped]

The starlite design is underrated.

 

[Cratty]

Keep in mind that while we set up our files in CMYK as requested, it’s unlikely that Tina is using CMYK printing. It is more likely 6-color printing, which would explain her ability to get very bright colors, especially in the blue range… and would also explain why certain colors print more as expected (red and greens) and some are a crapshoot (blues and purples).
Also, I would be careful when comparing the art to other members’ photos. The phone tends to brighten and saturate colors and they are often not accurate to the true chip color. I’ve had some Tina samples that looked very different when holding them, but look almost identical when photographed with my phone.

Tina may be able to exceed CMYK capabilities, but so long as our files are CMYK, any printing above should only be printer error/variance I would assume, right?

Color reading point is very fair , even with white/tone balancing there's still error introduced especially on phones (really need a spectrophotometer... used to have one at my old work.. miss it). Hope though is that it's somewhat consistent so that the overall trends identified can be somewhat consistent (E.g. magentas printing 5% higher or something).

But in the end what I want to somewhat understand is, is it even possible to correlate Tina's manufacturing error to a higher degree to determine what the colors will really look like. Seems like there's a chance for at least some tones to get some insights to get us closer. But hell, maybe just a fools errand. ha

 

[Colquhoun]

The i

Tina may be able to exceed CMYK capabilities, but so long as our files are CMYK, any printing above should only be printer error/variance I would assume, right?

Color reading point is very fair , even with white/tone balancing there's still error introduced especially on phones (really need a spectrophotometer... used to have one at my old work.. miss it). Hope though is that it's somewhat consistent so that the overall trends identified can be somewhat consistent (E.g. magentas printing 5% higher or something).

But in the end what I want to somewhat understand is, is it even possible to correlate Tina's manufacturing error to a higher degree to determine what the colors will really look like. Seems like there's a chance for at least some tones to get some insights to get us closer. But hell, maybe just a fools errand. ha

You’re right if she was taking a CMYK file and printing the file on a 6-color printer as CMYK, then it should be accurate to the file. I don’t know if she is converting them to 6-color, or RGB or what, but many colors get more saturated and bright, and some get lighter and a bit more washed-out.
I have been doing more a trial-and-error approach, and been getting samples of all chips I design. I’ve been avoiding the colors that don’t come out as I had envisioned, and repeatedly using colors I like. I’ve also purchased some samples of other members’ chips and asked if they’ll share the CMYK values of colors that I like.
I know it’s an inexact science, but it seems to be the most consistent way to get the colors I want.

 

[DeeVee8]

I was going to say you’ll probably need permission to recreate or use that mold design but then I remembered where we were buying from…

Well if it can’t be stopped then I would say let’s bring back the Proteges as a ceramic molded chip.

View attachment 794782
YOLO

Does this have a chance of being a thing? I'm in.