I was going to suggest Blackbloom rougue + Beledros witherbloom for black but that doesn't actually save a creature slot as the rouge would be non-expendable.
It looks like Cogwork archivist / epitah golem would just go infinite anyway with gyruda+broodmoth+strider if there are any even cmc enchantment creatures in the deck
I was going to suggest getting The ringheart quest into play off of the ultimatum for a spare green for roiling regrowth that can create a spare blue for mutating kinnan; but that doesn't work since the ringheart crest is the back side of a card.
The purpose of Cogwork archivist is to replace Quandrix command, but if it requires using Tyvar kell then it's not really an improvement. Also I'm not sure whether the bounce mode is adequately replaced.
Edit: though polukranos needs a black source too anyway so maybe it's ok
Hmm, I believe it might - mutating leosaur onto vadrok can put the island into the library, then paying a blue draws and plays the island, replaces the blue mana spent on the draw, and as a side effect puts some draw triggers on the stack. They can likely be used for profiting back the mana spent on the leosaur.
Also remember that cheap red doesn't work well with the back side of toralf.
Hmm; the other modes are used - the +1/+1 counter mode pumps Polukranos, unchained for fights, and the bounce mode is convenient in the opening; so they'd also need replacements. Those cards are also not compatible with openings based on Gyruda, doom of depths and Luminous broodmoth (and possibly just gyruda on its own), without which we'd need extra genesis ultimatums and/or wizard's spellbooks.
Ooh, nice! That gives two more creature slots.
Edit: We'll often have a lot of copies of legendaries for which the octopus is mutated on top, but Kargan intimidator can remove the octopus type from them.
So that leaves us with the the ringshart crest deck in #206 (ringheart crest is the back side of an mdfc, but the front side kolvori, god of kinship isn't problematic, and the back side can still be cast at instant speed from trickery as needed), with polukranos+swamp replacing primal might, for 9 layers (1 green -> X damage, 1 crest -> X green, 1 heliod's copy -> X crests (edit: 1 crest), 1 white -> X heliod's, 1 treasure -> X white, 1 room trigger -> X treasures, 4 ventures -> X room triggers, 1 combat -> X ventures, 1 main -> X combats, 2 mains) (or does that count as 8?); the alternative being the deck in #165 minus orb of dragonkind which is the same but 2 layers less.
So it looks like that deck is our best contender so far.
There's also Nylea's intervention + Angelic ascension (to give the opponent a flier) but that doesn't actually seem better than polukranos + swamp (copies from dual strike won't deal more damage); and the latter is one more creature for vault robber.
I also thought about Obosh to increase the damage dealt by each copy of Nylea's intervention, but Vadrok can be mutated onto Toralf to give it an odd cmc which would go infinite.
Even if the great henge is safe, which I'm not sure of, how does it help? If we kill all the nontoken creatures with toralf, then can replay them, this only seems to net as much blue mana as we have nontoken creatures, not a full stages worth.
Good point on maximising efficiency of the start; we do want to maximise treasures first (so having all the enchantment creatures in hand) and then optimise treasure usage second.
Best I can think of is after your start, play kinnan with one treasure, tap a treasure for UU and use to mutate octo + kinnan up skemfar, another 2 treasures for RRUU to vadrok up aracnoform and make a lot of kinnans + a few skemfars, use them to draw a few cards, then use 2 treasures for lots of U and R and continue as before. 6 treasures total.
The potential changes seem interesting, haven't checked them thoroughly. Bit nervous about making such big changes so close to the deadline though.
Another potential idea I've just had: Could we somehow use Aegar, the freezing flame to convert the output of the damage stage into input of the blue stage? If so that would free up life as an additional resource for later layers.
Something that life could be used for is that earlier there was an idea to use Orvar as the main copy engine instead of brooms, which would allow Avalanche caller to make a bunch of tapped snow lands for Beladros witherbloom to untap.
Oh, damn, just realised there's a problem in the current deck as well: Orb of dragonkind can produce extra mana doe to Kinnan, and that extra mana doesn't have the restriction of being only usable on dragons.
Have in the opening hand the two reflections and qcom. rather than more enchantment creatures; because nylea can draw those anyway.
Play kinnan, intimidator, then mutate octo onto kinnan.
Play a reflection on octo, making a lot of kinnans.
Tap two treasures for lots of U and lots of R. (5 treasures used total. I thought about using gyruda to fetch kinnan and intimidator to save 2 treasures; however that would result in fewer treasures in the first place so doesn't seem worth it)
Sac the original kinnan/octo
Mutate vadrok under something random, casting arachno to make a broom a copy of it. Play qcom from hand to bounce vadrok and shuffle octo. Sac what arachno is on.
Then, we can cast mutate vadrok under this new pile, playing arachnoform (making a broom a copy of it) and qcom (bouncing it), as many times as necessary (about 6). Also in between one of those steps we mutate octo onto faridah (that we fetch with kinnan; saving a red mana over casting it) and have brooms copy that too, and sac it, redrawing it with nalya after shuffling it with the next qcom. We could do the same with Toralf either now or later too. Finally play the investigator and we're good to go for the full stage.
(minor interaction I also noticed: once we have skemfar avenger in play; it would be convenient for it to never leave play; though creatures with arachnoform dying may cause it to trigger and lose us life when we don't want to. However, it's actually fine since we can either use intimidator to have the arachnoformd creature lose its types, or bounce the arachnoformed creature with qcom rather than saccing it).
Remember to take into account that gyruda mills 4 cards, so after at most 15 triggers we don't get anything more until we shuffle (ideally copying the qcommand a lot through stormcaller + dual strike)
Ah yes it looks like that works as a way to cut omen; and we don't need ETB brooms too often anyway (the stage works without making stormcaller every iteration, for instance)
(edit: Potential minor optimisation: in your start on #181, we don't really need xorn and robber so early; instead they could be fetched later by having a broom ETB as gyruda. Would it be better to get the thaumaturge out a little earlier? Or perhaps use a token maker to maximise early druid mana?)
Start: UUU, omen in grave, octo in hand island on field
Sac something, put broom triggers on the stack
Mutate octo (UU), 2 reflections
first: cast roiling regrowth, trickery octo, optional dual strike, qcom octo+omen+island+any 3 others
activate investigator; get draw triggers (U)
draw omen, get ETB brooms with it including cultivator (UU). This also makes the broom triggers still on the stack able to cause ETBs.
draw octo
resolve down to second octo; cast roiling regrowth, qcom island+any 2 others
resolve broom triggers for cultivator; get island (UUU)
back where we started with more brooms.
It also seems like a non-haste base would be tricky anyway. Other bases I'm aware of are paladium myr + kinnan (still needs haste), forsaken monument + crawling barrens (moots life, and requires animating an artifact which moots Orb of dragonkind), and strixhaven stadium + kinnan (which still requires animating an artifact to copy it), and birgi (not compatible with toralf)
Edit: Glasspool mimic/Umara Wizard + Beledros witherbloom + Azusa, Lost but Seeking looks like a way to get one extra blue mana from the opening, allowing for kinnan to be mutated and copied
It looks like Cogwork archivist / epitah golem would just go infinite anyway with gyruda+broodmoth+strider if there are any even cmc enchantment creatures in the deck
I was going to suggest getting The ringheart quest into play off of the ultimatum for a spare green for roiling regrowth that can create a spare blue for mutating kinnan; but that doesn't work since the ringheart crest is the back side of a card.
The purpose of Cogwork archivist is to replace Quandrix command, but if it requires using Tyvar kell then it's not really an improvement. Also I'm not sure whether the bounce mode is adequately replaced.
Edit: though polukranos needs a black source too anyway so maybe it's ok
Also remember that cheap red doesn't work well with the back side of toralf.
Edit: We'll often have a lot of copies of legendaries for which the octopus is mutated on top, but Kargan intimidator can remove the octopus type from them.
X crests(edit: 1 crest), 1 white -> X heliod's, 1 treasure -> X white, 1 room trigger -> X treasures, 4 ventures -> X room triggers, 1 combat -> X ventures, 1 main -> X combats, 2 mains) (or does that count as 8?); the alternative being the deck in #165 minus orb of dragonkind which is the same but 2 layers less.So it looks like that deck is our best contender so far.
There's also Nylea's intervention + Angelic ascension (to give the opponent a flier) but that doesn't actually seem better than polukranos + swamp (copies from dual strike won't deal more damage); and the latter is one more creature for vault robber.
I also thought about Obosh to increase the damage dealt by each copy of Nylea's intervention, but Vadrok can be mutated onto Toralf to give it an odd cmc which would go infinite.
Hmm, with the great henge then legendary creatures can always be killed cheaply just by cloning them. Does that go infinite?
Best I can think of is after your start, play kinnan with one treasure, tap a treasure for UU and use to mutate octo + kinnan up skemfar, another 2 treasures for RRUU to vadrok up aracnoform and make a lot of kinnans + a few skemfars, use them to draw a few cards, then use 2 treasures for lots of U and R and continue as before. 6 treasures total.
The potential changes seem interesting, haven't checked them thoroughly. Bit nervous about making such big changes so close to the deadline though.
Another potential idea I've just had: Could we somehow use Aegar, the freezing flame to convert the output of the damage stage into input of the blue stage? If so that would free up life as an additional resource for later layers.
Something that life could be used for is that earlier there was an idea to use Orvar as the main copy engine instead of brooms, which would allow Avalanche caller to make a bunch of tapped snow lands for Beladros witherbloom to untap.
Oh, damn, just realised there's a problem in the current deck as well: Orb of dragonkind can produce extra mana doe to Kinnan, and that extra mana doesn't have the restriction of being only usable on dragons.
Have in the opening hand the two reflections and qcom. rather than more enchantment creatures; because nylea can draw those anyway.
Play kinnan, intimidator, then mutate octo onto kinnan.
Play a reflection on octo, making a lot of kinnans.
Tap two treasures for lots of U and lots of R. (5 treasures used total. I thought about using gyruda to fetch kinnan and intimidator to save 2 treasures; however that would result in fewer treasures in the first place so doesn't seem worth it)
Sac the original kinnan/octo
Mutate vadrok under something random, casting arachno to make a broom a copy of it. Play qcom from hand to bounce vadrok and shuffle octo. Sac what arachno is on.
Then, we can cast mutate vadrok under this new pile, playing arachnoform (making a broom a copy of it) and qcom (bouncing it), as many times as necessary (about 6). Also in between one of those steps we mutate octo onto faridah (that we fetch with kinnan; saving a red mana over casting it) and have brooms copy that too, and sac it, redrawing it with nalya after shuffling it with the next qcom. We could do the same with Toralf either now or later too. Finally play the investigator and we're good to go for the full stage.
(minor interaction I also noticed: once we have skemfar avenger in play; it would be convenient for it to never leave play; though creatures with arachnoform dying may cause it to trigger and lose us life when we don't want to. However, it's actually fine since we can either use intimidator to have the arachnoformd creature lose its types, or bounce the arachnoformed creature with qcom rather than saccing it).
(edit: Potential minor optimisation: in your start on #181, we don't really need xorn and robber so early; instead they could be fetched later by having a broom ETB as gyruda. Would it be better to get the thaumaturge out a little earlier? Or perhaps use a token maker to maximise early druid mana?)
Start: UUU, omen in grave, octo in hand island on field
Sac something, put broom triggers on the stack
Mutate octo (UU), 2 reflections
first: cast roiling regrowth, trickery octo, optional dual strike, qcom octo+omen+island+any 3 others
activate investigator; get draw triggers (U)
draw omen, get ETB brooms with it including cultivator (UU). This also makes the broom triggers still on the stack able to cause ETBs.
draw octo
resolve down to second octo; cast roiling regrowth, qcom island+any 2 others
resolve broom triggers for cultivator; get island (UUU)
back where we started with more brooms.
It also seems like a non-haste base would be tricky anyway. Other bases I'm aware of are paladium myr + kinnan (still needs haste), forsaken monument + crawling barrens (moots life, and requires animating an artifact which moots Orb of dragonkind), and strixhaven stadium + kinnan (which still requires animating an artifact to copy it), and birgi (not compatible with toralf)