Photography with the DJI Mini 3 Pro - Vol 2

You are currently viewing Photography with the DJI Mini 3 Pro - Vol 2
"Little Planet" Panorama

RAW-processing with DxO PhotoLab 5.5

As I have already report­ed here, I had ordered a DJI Mini 3 Pro the day it was released. For a good 4 months now I have been using it, most recent­ly very inten­sive­ly dur­ing our vaca­tion in the Enga­dine in Switzer­land. The pic­ture above is a “Lit­tle Plan­et” panora­ma from Soglio in Bergell, more pic­tures are com­ing soon 😉.

As you can see from the oth­er arti­cles on my web­site, I’m pri­mar­i­ly inter­est­ed in pho­tog­ra­phy and use the drone pre­dom­i­nant­ly to take pic­tures. What par­tic­u­lar­ly attract­ed me to the Mini 3 Pro was its high-res­o­lu­tion 48 megapix­el sen­sor with a rel­a­tive­ly large sen­sor size of 1/1.3 inch­es - and that at a weight of less than 250g. This puts it in the drone cat­e­go­ry with the fewest restric­tions pos­si­ble, which great­ly expands its range of use.

In my last arti­cle about the DJI Mini 3 Pro, I had already test­ed its pho­to­graph­ic capa­bil­i­ties exten­sive­ly, also and espe­cial­ly in direct com­par­i­son with my DJI Air 2s and my Canon EOS R5. I was and still am pri­mar­i­ly inter­est­ed in the results of shoot­ing in RAW for­mat, which I use exclu­sive­ly in my pho­tog­ra­phy in order to achieve the max­i­mum pos­si­ble image quality.

In direct com­par­i­son with the 1 inch 20 megapix­el sen­sor of the DJI Air 2s, the DJ Mini 3 Pro with its 1/1.3 inch sen­sor with 48 megapix­els res­o­lu­tion showed over­all a high­er res­o­lu­tion. How­ev­er, the images also con­tained sig­nif­i­cant arti­facts in the 100% view, which were prob­a­bly due to an inter­nal noise reduc­tion of the DNG file on the one hand, and in par­tic­u­lar most like­ly due to the Quad-Bay­er for­mat used in the 1/1.3 inch sen­sor of the Mini 3 Pro (more info on this can be found here in my pre­vi­ous report).

Both of these are appar­ent­ly not yet opti­mal­ly sup­port­ed in the RAW work­flow I usu­al­ly use with Adobe Light­room Clas­sic* at the moment. Some­times, sur­pris­ing­ly, the JPG files cre­at­ed direct­ly in the drone showed a high­er res­o­lu­tion than the DNG files I processed in Lightroom.

So it should be pos­si­ble to get a bet­ter result with opti­mized software?

I had there­fore been wait­ing for a bet­ter suit­ed RAW con­vert­er. In par­tic­u­lar, I was hop­ing that DxO would be able to do this, as they had demon­strat­ed with Pho­to­Lab and Pur­eRAW what is pos­si­ble to get out of RAW files. I’ ve had very good results with both my Canon cam­eras and my pre­vi­ous DJI drones (the Mini 2 and the Air 2s).

And DxO has deliv­ered now:

DxO PhotoLab 5 supports the DJI Mini 3 Pro

Since the lat­est Pho­to­Lab ver­sion 5.5* (and also the lat­est ver­sion of Pur­eRAW 2), DxO now also sup­ports the DJI Mini 3 Pro.

I have there­fore reprocessed the sam­ple pic­tures of my DJI Mini 3 Pro from my pre­vi­ous report using DxO Pho­to­Lab 5.5 and would like to show the results here. The first image was devel­oped direct­ly in Light­room Clas­sic V 11.5, and the image below with the DxO Pho­to­Lab 5.5 plu­g­in. Both images were cre­at­ed from the same DNG file of the Mini 3 Pro:

DJI Mini 3 Pro DNG devel­oped with Light­room Classic
DJI Mini 3 Pro DNG devel­oped with DxO Pho­to­Lab 5.5 as Light­room plugin

By the way, I use DxO Pho­to­Lab 5 with the includ­ed Light­room plu­g­in only for RAW devel­op­ment and then edit the images in Adobe Light­room Clas­sic. After demo­saic­ing, denois­ing, and opti­cal cor­rec­tion, the plug-in sends them back to Light­room as DNG files, which allows me to per­form fur­ther exten­sive expo­sure cor­rec­tions. If you’re inter­est­ed in my rea­sons for doing this and my spe­cif­ic wok­flow, you can find details about it in my arti­cles about DxO Pho­to­Lab.

In the above overview images, which have been scaled down for web dis­play, no ground­break­ing changes are vis­i­ble yet, of course. There­fore, I now com­pare the areas that seemed prob­lem­at­ic to me pre­vi­ous­ly in high­er res­o­lu­tion. First, a 100% sec­tion of the cen­tral area:

Com­par­i­son DJI Mini 3 Pro DNG, left devel­oped in Light­room, right in DxO Pho­to­lab 5.5

In both pic­tures, I only adjust­ed the shad­ows in Light­room (Depths +100), the oth­er set­tings remained at their default val­ues. In direct com­par­i­son, I con­sid­er the DxO pro­cess­ing to be slight­ly rich­er in con­trast and detail. How­ev­er, com­pared to the shot with my Canon R5, the fine details are still miss­ing, which is espe­cial­ly notice­able in the bush­es. Below, I show the EOS R5 shot on the right for a direct comparison:

Com­par­i­son DJI Mini 3 Pro DxO vs Canon EOS R5 with RF 24-105 f/4L @ f/4

The blos­soms in the bush­es, which are clear­ly vis­i­ble as white dots in the pic­ture of the R5, can only just be sus­pect­ed in the pic­ture of the Mini 3 Pro. This is where the quad Bay­er design becomes notice­able again. The col­or res­o­lu­tion of the sen­sor is four times low­er than that of a clas­sic Bay­er sen­sor, so that small­est col­or details can­not be repro­duced. All in all, the DJI Mini 3 Pro does not do bad­ly. How­ev­er, I am pleased to say that my Canon EOS R5 still has its rai­son d’être.

I was par­tic­u­lar­ly curi­ous to see what the AI-based DxO Deep­PRIME algo­rithm could reveal in the depths of the Mini 3 Pro images. Due to the sig­nif­i­cant­ly small­er sen­sor size of the Mini 3 Pro com­pared to the full-frame cam­era, it nat­u­ral­ly pro­duces much more noise in dark areas and at high­er ISO values.

Here you can see again a com­par­i­son of a crop from my pre­vi­ous report. On the left, again in 100% res­o­lu­tion, you can see my pro­cess­ing direct­ly in Light­room Clas­sic, and on the right, the result of RAW devel­op­ment with Deep­PRIME in DxO Pho­to­Lab 5.5:

Com­par­i­son of RAW devel­op­ment in Light­room vs RAW devel­op­ment and Deep­PRIME denois­ing in DxO Pho­to­Lab 5.5

Here again, you can see how sen­sa­tion­al­ly effec­tive denois­ing is with DxO’s Deep­PRIME algo­rithm. Even more impres­sive­ly, you can see the high-qual­i­ty of the noise reduc­tion in the sky. Here’s a detail from the over­all shot with the church spire, this time even at 200% mag­ni­fi­ca­tion. The fine noise in the blue of the sky is com­plete­ly removed by Pho­to­Lab 5.5 with the Deep­PRIME algo­rithm, with­out los­ing any of the detail in the spire:

Com­par­i­son of RAW devel­op­ment in Light­room vs RAW devel­op­ment and Deep­PRIME denois­ing in DxO Pho­to­Lab 5.5, 200% crop

And for com­par­i­son, here is the same detail again in com­par­i­son with my Canon EOS R5 (in the pic­ture on the right):

Church spire, left DJI Mini 3 Pro devel­oped with DxO, right EOS R5 with RF 24-105 f/4L, 200% Crop

Here, the dif­fer­ence to the full-frame cam­era is bare­ly vis­i­ble. All in all, the results of RAW devel­op­ment in DxO Pho­to­Lab 5.5 are once again con­vinc­ing. Although the res­o­lu­tion of details in the images from my DJI Mini 3 Pro is still low­er than that of my full-frame sys­tem cam­era. That was to be expect­ed. Still, the results are sur­pris­ing­ly good for a drone that weighs less than half as much as the body of my Canon EOS R5 alone. Thanks to DxO Pho­to­Lab 5.5, I can now expect even bet­ter results with it.

By the way, DxO offers the great oppor­tu­ni­ty to put the pro­grams through their paces as unlim­it­ed tri­al ver­sions for 30 days. They can be down­loaded from the fol­low­ing links:

DxO Pur­eRAW 2 tri­al ver­sion*

Pho­to­Lab 5 Elite tri­al ver­sion*

I rec­om­mend every­one to see for them­selves what is pos­si­ble with Deep­PRIME. I have already been able to make some old noisy pic­tures shine in new bril­liance with it.

* = Affil­i­ate link

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.