Testing and evaluating KODAK T-MAX P3200 and Ilford Delta 3200

[aparat]

The Kodak T-MAX P3200 and Ilford Delta 3200 are highly popular and highly respected panchromatic films with high speed, respectably small grain, and beautiful tonality. I thought perhaps we could discuss their performance in this thread. I am going to perform a few different tests along the way and report my results once they become available. I hope you guys will offer your own comments and suggestions of what and how else to test these two films.

to start with, I ran a conventional curve-family test of both films, giving them the same exposure and processing them in the same tank. I ran the full test twice, each time obtaining very similar results. I then averaged density readings over the two trials. There is a lot of data to go through, so I am going to present it in small chunks, staring with a summary and curve family plots for each film. There's a lot more detail coming, hopefully, as the thread continues. At the end of my analysis, I will publish a detailed report.

The Kodak curves turned out a bit lumpy. I used two different rolls (both fresh), coming from two different mail orders to make sure I did not accidentally end up with a defective product. In each case, the curves ended up looking somewhat lumpy, especially compared to the well-formed Ilford curves. The two films have a different response to exposure and development. Kodak has a long toe and nicely controlled mid-tone and highlight contrast, as the development time increases. The film actually gains a bit of speed (shadow detail) over the range of processing times, of almost a whole stop, whereas Delta 3200 produces virtually no gain in speed, only in contrast. Bear in mind that I used D76, so the results would probably be different with DDX and XTOL. I am planning to use XTOL to push-process these films in the near future. Oh, and you will also notice that the Kodak film has a higher B+F density, at least with D76.

I plotted the curves using absolute log exposure values and calculating ISO film speed based on them. There's a summary table included with each plot. I must emphasize that I used an uncalibrated, DIY sensitometer, so the results are meant to be rough, "ballpark" estimates, only. I do have a calibrated and certified sensitometer, but decided to use the same exact setup as I did for my recent test of the CatLABS film, just to keep it consistent. I gave the films an exposure of around 2.24 MLS, which I estimated to be approximately what's needed for an ISO 1000 film. Again, there's a lot more detail coming soon.

 

[aparat]

I happen to really like Phil Davis's Beyond the Zone System method and I think that a lot of photographers are familiar with it. So this time, I am including an analysis of the Kodak T-MAX P3200 curves done according to the BTZS method, which also happens to be what Win Plotter does. Win Plotter is an excellent, proprietary application for MS Windows. Instead of Win Plotter, I am using a cross-platform application that I am currently working on, which, hopefully, I will finish at some point and make available free-of-charge.

The first plot contains a summary of the curve family analysis. Instead of using actual log exposure values, the BTZS method recommends using a "Personal Speed Point" (PSP) and calculating all of the relevant parameters (such as film speed, Ḡ, CI, Gamma, etc.) based on the position of each curve relative to the PSP. In theory, this type of analysis should produce results consistent with the previous analysis I posted, i.e., using absolute log exposure values. Note the two curves marked with a dotted line ("ISO 12 min." and "L7 9.24 min.). More on them down below.



The second plot shows the theoretical curve that complies with the ISO standard for measuring film speed, i.e., a curve with the Average Gradient, Ḡ = 0.62. In a conventional curve family test, such a curve is likely to land between two measured curves - in this case between the 11 min. and the 13.5 min. curves. I added a Zone System frame of reference to the plot, showing the seven-zone range that a lot of photographers consider important toward producing fine prints and easy scans.



The third plot shows another theoretical curve, this time one that describes the Subject Luminance Range (SLR) of seven stops (Phil Davis often uses the term "Subject Brightness Range, which is more of a perceptual descriptor"). The SLR of seven stops is generally considered to describe a "standard" outdoor scene with clearly defined shadows, mid-tones, and highlights, spanning approximately 7 EV. This curve is perhaps less useful for the EI 3200 films, such as the Kodak T-MAX P3200, which aren't likely to used outdoors in good light. Still, I think it's an interesting way to think about film performance. Ideally, the curve would offer perfectly linear performance across the entire range. The P3200 curve does this fairly well. To capture such a scene, one would need to expose the film for around EI 800+ and develop in D76 stock for around 9:30 min. at 20C (in a rotary processor).

 

[Paul Howell]

Interesting, I do shoot Tmax 3200 in good lighting when using a long lens and I want to keep the shutter speed at 1/2000 or higher and a F stop of F16 or 22. In the past I've used Tmax and DDX developer, current is Clayson F76+ which I think has many of the attributes of D76. Based on your findings if shooting in good light without a need for the higher ISO I should just shoot Tmax 400 at 800, although Kodak advises to use the same development time as at 400 maybe give it 10% more.

 

[Sirius Glass]

I am looking forward to see your results and comments.

 

[aparat]

Interesting, I do shoot Tmax 3200 in good lighting when using a long lens and I want to keep the shutter speed at 1/2000 or higher and a F stop of F16 or 22. In the past I've used Tmax and DDX developer, current is Clayson F76+ which I think has many of the attributes of D76. Based on your findings if shooting in good light without a need for the higher ISO I should just shoot Tmax 400 at 800, although Kodak advises to use the same development time as at 400 maybe give it 10% more.

You're right. I haven't thought about the use case you describe. I think that you would be okay shooting, as you suggest, T-MAX 400 at EI 800 and push processing it. I only have data from the older TMY emulsion, but I have no reason to suspect the new one is any slower. Here's a theoretical ISO curve for KODAK T-MAX 400 (TMY) in D76 1+1. It's just a hair below ISO 400, with a very nicely linear response, but in DDX or Clayton F76+ it might be even faster.

Is the Clayton F76+ a kind of "clone" of DDX? I used it quite a lot in the past and remember it very fondly.

 

[Andrew O'Neill]

What software are you using to create your curves?

 

[Sirius Glass]

I am looking forward to see your results and comments.

I am looking to using those films for the 500mm Hasselblad lens with a 2XE extender at smaller f/stops for better depths of field.

 

[aparat]

@Andrew O'Neill I wrote the software, or, more precisely, I am still writing it. I've gotten a lot of inspiration and learned a lot from the Photrio community over the years. When it's done, I will make it available for free. Currently, it uses a command line interface and reads data from text files, but I am hoping to build a UI to work in a web browser, with some data entry and editing capability. I still have a long way go.

@Sirius Glass That's a great use case I haven't thought of. The longest MF lens I own is 180mm, so it never occurred to me that other photographers might use much longer lenses.

 

[ic-racer]

Two things, did you mention on the other thread which sensitomter you were using and where it was calibrated? I don't recall.

Also, could you post the dataset, I'd like to compare the 'automated' results with my automated Excel spreadsheet.

Thanks!

 

[Andrew O'Neill]

@Andrew O'Neill I wrote the software, or, more precisely, I am still writing it. I've gotten a lot of inspiration and learned a lot from the Photrio community over the years. When it's done, I will make it available for free. Currently, it uses a command line interface and reads data from text files, but I am hoping to build a UI to work in a web browser, with some data entry and editing capability. I still have a long way go.

@Sirius Glass That's a great use case I haven't thought of. The longest MF lens I own is 180mm, so it never occurred to me that other photographers might use much longer lenses.

That's fantastic news, aparat! What's the timeline?

 

[aparat]

That's fantastic news, aparat! What's the timeline?

Thanks! It'll be a while before I have anything worth sharing. If all goes well, maybe sometime around February 2023.

@ic-racer I sent you a DM with the details you asked about. Please, let me know if there's anything else.

 

[DREW WILEY]

Neither is even remotely that high a speed given ordinary rating standards. Read the "fine print" in the respective tech spec sheets, more like around 1000. But they have a curve structure allowing you to salvage something or other if shot at higher speeds. I prefer 800 for either, which is only a stop faster than TMY400 with its far finer grain. I'm not barking that one is better than another, even if all three are shot at 800 and souped in the same developer. They just all have very different personalities and detail rendering. TMY is almost in a class by itself by combining very fine grain and high resolution with high speed; but it's more contrasty than the other two.

Speed-wise and curve shape, old and current TMY's are just about the same; but the sorta grain clumping characteristic of the first version was ironed out in the second, so a distinct improvement. I prefer PMK staining pyro for all the above.

 

[Paul Howell]

Looking at Kodak's data sheet, Tmax 3200 can be shot at 3200 when developed in Tmax developer, as noted by OP it does result in a high contrast negative, but printable. I agree that in most situations Tmax 400 is a better film.

Is the Clayton F76+ a kind of "clone" of DDX? I used it quite a lot in the past and remember it very fondly.

I think F76+ is closer to HC110, not an exact match of course, good balance of grain, contrast, speed and apparent sharpness. The closest to TMAX developer was Clayton F90, no longer in production.

 

[aparat]

The more I work with the T-MAX P3200, the more I am impressed with it. I think this emulsion is engineered to give a real speed increase with increased development.



I am curious how it will respond to more aggressive underexposure (EI 3200) and processing with XTOL. That's what I am planning to look at next.

By the way, I made a small error in my earlier calculations by including the density of a glass holder that's part of my setup twice. The estimated speed of the KODAK T-MAX P3200 is around ISO 1080. Sorry about that.

 

[warden]

I'll never use a CatLabs product but have enjoyed your analysis of their film in the other threads. But these two films I use fairly frequently and intend to continue doing so, so I'm keen to learn what practical takeaways there may be from this analysis. Thanks in advance!

 

[aparat]

I think a lot of film photographers would agree that the ultimate product in photography is a fine print. Lots of criteria exist regarding what is and what is not a printable negative. Curve family and tone reproduction analyses can, to some degree, predict how negatives would print. I say "to some degree" because skilled printers can work near-miracles from badly exposed and developed negatives, not to mention the general lack of agreement of what makes a fine print. So please, take the plots below with a grain of salt.

The 13.5 min. curve (approx. EFS 1250−) of the KODAK T-MAX P3200, according to the data, seems barely printable on Grade 2 paper (Ilford MGIV FB Glossy) with a condenser enlarger and appears to be more easily printable with diffusion enlarger (or by contact printing).

Here's a curve plot of the 13.5 min. curve showing a lot of data outside of the crucial seven-zone range. Significant dodging and burning would be required.


And here's a different look at the same data, showing tone reproduction analysis of the same curve, this time, simulated with a diffusion enlarger (or contact print) and flare-resistant camera and enlarger lenses. Here, the negative is somewhat more easily printable, with some inevitable tonal compression in the shadows and highlights.

 

[Stephen Benskin]

I think a lot of film photographers would agree that the ultimate product in photography is a fine print. Lots of criteria exist regarding what is and what is not a printable negative. Curve family and tone reproduction analyses can, to some degree, predict how negatives would print. I say "to some degree" because skilled printers can work near-miracles from badly exposed and developed negatives, not to mention the general lack of agreement of what makes a fine print. So please, take the plots below with a grain of salt.

The 13.5 min. curve (approx. EFS 1250−) of the KODAK T-MAX P3200, according to the data, seems barely printable on Grade 2 paper (Ilford MGIV FB Glossy) with a condenser enlarger and appears to be more easily printable with diffusion enlarger (or by contact printing).

Here's a curve plot of the 13.5 min. curve showing a lot of data outside of the crucial seven-zone range. Significant dodging and burning would be required.
View attachment 322236

And here's a different look at the same data, showing tone reproduction analysis of the same curve, this time, simulated with a diffusion enlarger (or contact print) and flare-resistant camera and enlarger lenses. Here, the negative is somewhat more easily printable, with some inevitable tonal compression in the shadows and highlights.

View attachment 322237

I wrote my own plotting program because I believe Phil Davis' version has too many compromises in order to bring it into agreement with BTZS. He over-simplified the camera image and tone reproduction curves. I was especially frustrated with how he calculated and depicted flare. Adding flare to the film's characteristic curve simplifies the process in that it can eliminate the need for a camera image curve. While the resulting density values can be accurately determined, I believe you lose the impression on how flare actually works. That's why I like the four quadrant reproduction curve.

In the above example, some of the numbers don't quite fit if the example is supposed to be for a normal or statistically average situation. I'm uncertain whether the flare value of 0.5 means 1/2 or is in log units. The NDR and CI is too high for a Subject Luminance Range of 2.20 to fit on a grade two diffusion. Also an LER of 1.29 is outside of what is considered grade two. It appears to be attempting to fit Zone System aims into a sensitometric model.

You might want to check out Exposure-Speed Relations and Tone Reproduction by Jack Holm. Be sure to treat his Zone System values with a little skepticism. He uses 128% for the highlight Reflectance instead of 100%.

http://beefalobill.com/benskin/content/Speed%20Relations%20and%20Tone%20Reproduction%20-%2005.03.2014.pdf

I put something together some time ago that walks though the photographic process using the equations and plugging in values.

http://beefalobill.com/benskin/content/Defining%20K.pdf

And something Davis never applied, here is something I wrote on the Delta-X Criterion.

 

[Paul Howell]

You need to remember that Tmax 3200 was designed for low light work, surveillance, news, PJism, shadow detail was not a primary consideration. To understand the capacity of the film you need to use DDX, Tmax developer or Acufine. I used Acufine for a couple of rolls, was quite pleased, then Kodak pulled Tmax 3200 from the market so did not follow up.

 

[DREW WILEY]

I don't lament the loss of TMZ. TMY is not only a lot more versatile, but comes in all standard film sizes.

 

[aparat]

I wrote my own plotting program because I believe Phil Davis' version has too many compromises in order to bring it into agreement with BTZS. He over-simplified the camera image and tone reproduction curves. I was especially frustrated with how he calculated and depicted flare. Adding flare to the film's characteristic curve simplifies the process in that it can eliminate the need for a camera image curve. While the resulting density values can be accurately determined, I believe you lose the impression on how flare actually works. That's why I like the four quadrant reproduction curve.

In the above example, some of the numbers don't quite fit if the example is supposed to be for a normal or statistically average situation. I'm uncertain whether the flare value of 0.5 means 1/2 or is in log units. The NDR and CI is too high for a Subject Luminance Range of 2.20 to fit on a grade two diffusion. Also an LER of 1.29 is outside of what is considered grade two. It appears to be attempting to fit Zone System aims into a sensitometric model.

You might want to check out Exposure-Speed Relations and Tone Reproduction by Jack Holm. Be sure to treat his Zone System values with a little skepticism. He uses 128% for the highlight Reflectance instead of 100%.

http://beefalobill.com/benskin/content/Speed%20Relations%20and%20Tone%20Reproduction%20-%2005.03.2014.pdf

I put something together some time ago that walks though the photographic process using the equations and plugging in values.

http://beefalobill.com/benskin/content/Defining%20K.pdf

And something Davis never applied, here is something I wrote on the Delta-X Criterion.

Thank you for the comments and the references! They're much appreciated. I will respond briefly, as I do not want to wander off topic, but I would be interested in discussing this in a dedicated thread, if you are willing to participate, or via DM.

Much of the knowledge resides in old journals and among enthusiasts/experts, like yourself. I do agree about Phil Davis' BTZS but I also appreciate the practical application of theory that he included in his book.

Regarding flare, I have implemented a few different methods and models so far. For example, the Phil Davis method based on the notion of "flare density" is what perhaps most photographers who read BTZS are familiar with. What I think you're referring to is the flare calculation in Quadrant 1, which is based on a "flare factor," as in Hf = (Hs * 2^F)-Hs, which is added to the camera image exposure at the film plane. Here, the flare factor, F, is in stops.

I agree that the CI is too high for Grade 2, but I wanted to include the example simulating a situation where a skilled printer can make that work. The P3200 is a film that most people, it seems, are going to push, so they're likely dealing with CI of over 0.7 most of the time.

As far as the paper data, I did this test a long time ago, so it may not be the most accurate or relevant. I should have mentioned that. I haven't yet tested the new Ilford MG emulsion, but I hope to be able to do that at some point.

Finally, you are absolutely right about mixing Zone System and sensitometric models. I haven't yet settled on how I want to do this. I want to find a way to present sensitometric data in a way that makes sense to film photography practitioners, many of whom use the Zone System. I need some advice and feedback from photographers regarding this issue because, frankly, I am stuck. Here's how Phil Davis did this in Win Plotter, as a quick example. It's not my favorite because I don't think it's very effective in conveying the message.

 

[aparat]

You need to remember that Tmax 3200 was designed for low light work, surveillance, news, PJism, shadow detail was not a primary consideration. To understand the capacity of the film you need to use DDX, Tmax developer or Acufine. I used Acufine for a couple of rolls, was quite pleased, then Kodak pulled Tmax 3200 from the market so did not follow up.

Yes, DDX is on my radar. I currently only have XTOL out of those recommended by Kodak for pushing P3200. Do you happen to know if XTOL is going to do the job as well as DDX?

 

[gone]

F76 is my go to developer for a lot of different films. It's sort of a "cleaner" developer than D76, for my purposes anyway. That of course depends on how it's diluted and agitated. I give it a fair amount of agitation and love the results.

I like liquid developers these days, just easier to mix up. The F76 will last quite a while after being opened up too, and still look about the same for quite a while in the bottle. Very consistent. D76 always looked good at first, then got stranger once it was mixed up and sat for a while. 2 months was all I ever got before that happened. Using it full strength, it was used up pretty quickly though.

 

[Paul Howell]

I have not used Xtol with Tmax 3200, as a matter of fact have not used Xtol for a very long time, had a gallon of the stuff die the sudden death, then Kodak stopped making the 1 liter or quart size. Looking at the Kodak data sheet it seems that best for push is Tmax developer, Kodak lists 6400 as a push for TMX developer, but Xtol seems to a good choice as well, but is 3200 a one or two stop push? What I recall from using Tmax developer and DDX is that 3200 is not a push, which is why Kodak DX codes Tmax 3200 at 3200 and recommends Tmax developer. I found no real difference between TMAX and DDX. Different formula but I think Aucfine is a better choice for push. I just bought a quart size Acufine, I have a few cans of replenisher.

 

[Sirius Glass]

I have not used Xtol with Tmax 3200, as a matter of fact have not used Xtol for a very long time, had a gallon of the stuff die the sudden death, then Kodak stopped making the 1 liter or quart size. Looking at the Kodak data sheet it seems that best for push is Tmax developer, Kodak lists 6400 as a push for TMX developer, but Xtol seems to a good choice as well, but is 3200 a one or two stop push? What I recall from using Tmax developer and DDX is that 3200 is not a push, which is why Kodak DX codes Tmax 3200 at 3200 and recommends Tmax developer. I found no real difference between TMAX and DDX. Different formula but I think Aucfine is a better choice for push. I just bought a quart size Acufine, I have a few cans of replenisher.

But I am interested in Tmax 3200 and Ilford Delta 3200 both @Iso 3200developed in XTOL and replenished XTOL. That is where the rubber meets the road.

 

[Stephen Benskin]

Thank you for the comments and the references! They're much appreciated. I will respond briefly, as I do not want to wander off topic, but I would be interested in discussing this in a dedicated thread, if you are willing to participate, or via DM.

Much of the knowledge resides in old journals and among enthusiasts/experts, like yourself. I do agree about Phil Davis' BTZS but I also appreciate the practical application of theory that he included in his book.

Regarding flare, I have implemented a few different methods and models so far. For example, the Phil Davis method based on the notion of "flare density" is what perhaps most photographers who read BTZS are familiar with. What I think you're referring to is the flare calculation in Quadrant 1, which is based on a "flare factor," as in Hf = (Hs * 2^F)-Hs, which is added to the camera image exposure at the film plane. Here, the flare factor, F, is in stops.

I agree that the CI is too high for Grade 2, but I wanted to include the example simulating a situation where a skilled printer can make that work. The P3200 is a film that most people, it seems, are going to push, so they're likely dealing with CI of over 0.7 most of the time.

As far as the paper data, I did this test a long time ago, so it may not be the most accurate or relevant. I should have mentioned that. I haven't yet tested the new Ilford MG emulsion, but I hope to be able to do that at some point.

Finally, you are absolutely right about mixing Zone System and sensitometric models. I haven't yet settled on how I want to do this. I want to find a way to present sensitometric data in a way that makes sense to film photography practitioners, many of whom use the Zone System. I need some advice and feedback from photographers regarding this issue because, frankly, I am stuck. Here's how Phil Davis did this in Win Plotter, as a quick example. It's not my favorite because I don't think it's very effective in conveying the message.

View attachment 322248

You're calculating flare correctly. What confused me was how you are noting it.

A flare factor of 1 stop would be noted as 2. Based on that, a 0.5 is really small. 0.5 could mean 1/2 stop. Or it could mean 0.5 logs or 1 2/3 stops. The degree of flare is not exactly clear from the notation, but from the value of HR, it shows flare reduced the illuminance range by 1/2 stop.

Flare for an average scene is considered to be around 1 stop. Since the number of lens elements has an affect on flare, the more elements, the higher the flare. Kodak went from 1 to 1 1/3 stops.. My belief is because of the greater use of 35 mm cameras over view cameras. Consider the equation on how to determine the aim average gradient for a given luminance range. Kodak considers 0.58 to be for a statistically average 2.20 scene. For an LER of 1.05 (grade 2 diffusion), simply dividing 2.20 into 1.05 would produce an average gradient of 0.48. While the scene may be 2.20, the illuminance range striking the film is more around 1.90 (one stop flare) or 1.80 (1 1/3 stops).

1.05 / (2.20 - 0.4) = 0.58

With your conditions as stated, the aim CI is 0.51 based on the flare factor. For a CI of 0.70 with a stop flare, the illuminance range needs to be 1.50, which is about what the TMZ negative would lose when rated at EI 3200 (considering a lot of caveats).

The basic idea of "pushing for speed" is to increase the gradient to compensate for the lower density range due to the loss of shadow detail from underexposure. If done correctly, the negative should have the same NDR and fit the same LER as a correctly exposed negative with normal development. I believe in your example, the camera image doesn't reflect the underexposure. If it did, the shadow would be placed lower on the curve. This would reduce the effective NDR. Here's an example of underexposure.

Compare the reproduction curve to an "ideal" Normal.

For paper, the LER doesn't mean black to white. From a Zone System standpoint, it's more from Zone 1 to 1.5 to Zone 8 to 9. There's still room on the paper.

A good tone reproduction curve is one where the middle gradient is above 1.08-1.10 and the curve falls below the guide curve (reproduction tones are lighter than the original). People can accept a certain amount of compression of the shadows and highlights in the reproduction, but the mid-tones almost always need to exceed the aim number.

What I did was to research what the values should be for a statistically average scene. I call this the standard model. I used these numbers to test the program while developing it. I believe these are included in the Defining K paper.